Ibm OPEN STORAGE READ ME FIRST 7-9-2010, I VIRTUALIZATION READ ME FIRST 7-9-2010 User Manual

Download

IBM i Virtualization and Open Storage Read-me First

Vess Natchev

Cloud | Virtualization | Power Systems

IBM Rochester, MN

vess@us.ibm.com

July 9th, 2010

This “read-me first” document provides detailed instructions on using IBM i 6.1 virtualization and connecting open storage to IBM i. It covers prerequisites, supported hardware and software, planning considerations, install and post-install tasks such as backups. The document also contains links to many additional information sources.

Table of Contents

1. IBM i virtualization solutions

1.1. IBM i logical partition (LPAR) hosting another IBM i partition

1.2. IBM i using open storage as a client of the Virtual I/O Server (VIOS)

1.3. IBM i on a Power blade

2. IBM i hosting IBM i supported configurations

2.1. Hardware

2.2. Software and firmware

3. IBM i hosting IBM i concepts

3.1. Virtual SCSI and Ethernet adapters

3.2. Storage virtualization

3.3. Optical virtualization

3.4. Network virtualization

4. Prerequisites for implementing IBM i hosted LPARs

4.1. Storage planning

4.2. Performance

4.3. Dual hosting

5. Implementing IBM i client LPARs with an IBM i host

6. Post-install tasks and considerations

6.1. Configure IBM i networking

6.2. How to perform IBM i operator panel functions

6.3. How to display the IBM i partition System Reference Code (SRC) history

6.4. Client IBM i LPARs considerations and limitations

6.5. Configuring Electronic Customer Support (ECS) over LAN

6.6. Copying storage spaces

6.7. Backups

7. IBM i using open storage supported configurations

7.1. Hardware

7.2. Software and firmware

8. IBM i using open storage through VIOS concepts

8.1. Virtual SCSI and Ethernet adapters

8.2. Storage virtualization

8.3. Optical virtualization

8.4. Network virtualization

9. Prerequisites for attaching open storage to IBM i through VIOS

9.1. Storage planning

9.2. Performance

9.3. Dual hosting and multi-path I/O (MPIO)

9.3.1. Dual VIOS LPARs with IBM i mirroring

9.3.2. Path redundancy to a single set of LUNs

9.3.2.1. Path redundancy with a single VIOS

9.3.2.2. Redundant VIOS LPARs with client-side MPIO

9.3.3 Subsystem Device Driver – Path Control Module (SDDPCM)

10. Attaching open storage to IBM i through VIOS

10.1. Open storage configuration

10.2. VIOS installation and configuration

10.3. IBM i installation and configuration

10.4. HMC provisioning of open storage in VIOS

10.5. End-to-end LUN device mapping

11. Post-install tasks and considerations

11.1. Configure IBM i networking

11.2. How to perform IBM i operator panel functions

11.3. How to display the IBM i partition System Reference Code (SRC) history

11.4. Client IBM i LPARs considerations and limitations

11.5. Configuring Electronic Customer Support (ECS) over LAN

11.6. Backups

12. DS4000 and DS5000 Copy Services and IBM i

12.1. FlashCopy and VolumeCopy

12.2. Enhanced Remote Mirroring (ERM)

13. IBM i using SAN Volume Controller (SVC) storage through VIOS

13.1. IBM i and SVC concepts

13.2. Attaching SVC storage to IBM i

14. SVC Copy Services and IBM i

14.1. FlashCopy

14.1.1. Test scenario

14.1.2. FlashCopy statements

14.2. Metro and Global Mirror

14.2.1. Test scenario

14.2.2. Metro and Global Mirror support statements

15. XIV Copy Services and IBM i

16. DS5000 direct attachment to IBM i

16.1. Overview

16.2. Supported hardware and software

16.3. Best practices, limitations and performance

16.4. Sizing and configuration

16.5. Copy Services support

17. N_Port ID Virtualization (NPIV) for IBM i

17.1. Overview

17.2. Supported hardware and software

17.3. Configuration

17.4. Copy Services support

18. Additional resources

19. Trademarks and disclaimers

1. IBM i virtualization solutions

IBM i 6.1 introduces three significant virtualization capabilities that allow faster deployment of IBM i workloads within a larger heterogeneous IT environment. This section will introduce and differentiate these new technologies.

1.1. IBM i logical partition (LPAR) hosting another IBM i partition

An IBM i 6.1 LPAR can host one or more additional IBM i LPARs, known as virtual client LPARs. Virtual client partitions can have no physical I/O hardware assigned and instead leverage virtual I/O resources from the host IBM i partition. The types of hardware resources that can be virtualized by the host LPAR are disk, optical and networking. The capability of IBM i to provide virtual I/O resources has been used successfully for several years to integrate AIX®, Linux® and Windows® workloads on the same platform. The same virtualization technology, which is part of the IBM i operating system, can now be used to host IBM i LPARs. IBM i hosting IBM i is the focus of the first half of this document.

1.2. IBM i using open storage as a client of the Virtual I/O Server (VIOS)

IBM i virtual client partitions can also be hosted by VIOS. VIOS is virtualization software that runs in a separate partition whose purpose is to provide virtual storage, optical, tape and networking resources to one or more client partitions. The most immediate benefit VIOS brings to an IBM i client partition is the ability to expand its storage portfolio to use 512-byte/sector open storage. Open storage volumes (or logical units, LUNs) are physically attached to VIOS via a Fibre Channel or Serial-attached SCSI (SAS) connection and then made available to IBM i. While IBM i does not directly attach to the SAN in this case, once open storage LUNs become available through VIOS, they are managed the same way as integrated disks or LUNs from a directly attached storage system. IBM i using open storage through VIOS is the focus of the second half of this read-me first guide.

1.3. IBM i on a Power blade

The third major virtualization enhancement with IBM i 6.1 is the ability to run an IBM i LPAR and its applications on a Power blade server, such as IBM BladeCenter JS12 or JS22. Running IBM i on a Power blade is beyond the scope of this document. See the IBM i on a Power Blade Read- me First for a complete technical overview and implementation instructions: http://www.ibm.com/systems/power/hardware/blades/ibmi.html

2. IBM i hosting IBM i supported configurations

2.1. Hardware

One of the most significant benefits of this solution is the broad hardware support. Any storage, network and optical adapters and devices supported by the host IBM i partition on a POWER6 processor-based server can be virtualized to the client IBM i partition. Virtualization of tape devices from an IBM i host to an IBM i client is not supported. The following table lists the supported hardware:

Hardware type Supported for

IBM i hosting

IBM i

IBM Power servers Yes Includes IBM Power 520 Express, IBM

Power 550 Express, IBM Power 560 Express, IBM Power 570 and IBM Power 595

Does not include IBM POWER6 processorbased blade servers, such as IBM

BladeCenter JS12 and JS22 IBM Power 575 No POWER5-based systems or earlier Storage adapters (Fibre Channel, SAS, SCSI)

Storage devices and subsystems

Network adapters Yes Must be supported by IBM i 6.1 or later and

Optical devices Yes Must be supported by IBM i 6.1 or later and

Tape devices No To determine the storage, network and optical devices supported on each IBM Power server

model, refer to the Sales Manual for each model: http://www.ibm.com/common/ssi/index.wss To determine the storage, network and optical devices supported only by IBM i 6.1, refer to the

upgrade planning Web site: https://www-

304.ibm.com/systems/support/i/planning/upgrade/futurehdwr.html.

2.2. Software and firmware

Software or firmware type Supported for

IBM i 6.1 or later Yes Required on both host and client IBM i IBM i 5.4 or earlier No Not supported on host or client partition

IBM Power server system firmware 320_040_031 or later HMC firmware HMC V7 R3.2.0 or later

3. IBM i hosting IBM i concepts

The capability of an IBM i partition to host another IBM i partition involves hardware and virtualization components. The hardware components are the storage, optical and network adapters and devices physically assigned to the host IBM i LPAR. The virtualization components

Yes Must be supported by IBM i 6.1 or later and

supported on POWER6-based IBM Power

server

Yes Must be supported by IBM i 6.1 or later and

supported on POWER6-based IBM Power

server

supported on POWER6-based IBM Power

server

supported on POWER6-based IBM Power

server

IBM i hosting

IBM i

partition

Yes This is the minimum system firmware level

required

Yes This is the minimum HMC firmware level

required

Notes

are the system firmware and IBM i operating system objects necessary to virtualize the physical I/O resources to client partitions. The following diagram shows the full solution and its components:

3.1. Virtual SCSI and Ethernet adapters

IBM i hosting IBM i uses an existing function of the system firmware, or Power Hypervisor: the capability to create virtual SCSI and Ethernet adapters in a partition. Virtual adapters are created for each LPAR in the Hardware Management Console (HMC). Virtual SCSI adapters are used for storage and optical virtualization; virtual Ethernet adapters are used for network virtualization.

Note that using virtual I/O resources from a host partition does not preclude an IBM i client partition from owning physical hardware. A mix of virtual and physical hardware in the same partition is supported for IBM i in this environment, by assigning both types of adapters to the partition in the HMC.

3.2. Storage virtualization

To virtualize integrated disk (SCSI, SAS or SSD) or LUNs from a SAN system to an IBM i client partition, both HMC and IBM i objects must be created. In the HMC, the minimum required configuration is:

• One virtual SCSI server adapter in the host partition

• One virtual SCSI client adapter in the client partition

This virtual SCSI adapter pair allows the client partition to send read and write I/O operations to the host partition. More than one virtual SCSI pair can exist for the same client partition in this environment. To minimize performance overhead on the host partition, the virtual SCSI connection is used to send I/O requests, but not for the actual transfer of data. Using the capability of the Power Hypervisor for Logical Remote Direct Memory Access (LRDMA), data are transferred directly from the physical adapter assigned to the host partition to a buffer in memory of the client partition.

There is no additional configuration required in IBM i in the virtual client partition. In the host partition, the minimum required IBM i setup consists of the following:

• One Network Server Description (NWSD) object

• One Network Server Storage Space (NWSSTG) object

The NWSD object associates a virtual SCSI server adapter in IBM i (which in turn is connected to a virtual SCSI client adapter in the HMC) with one or more NWSSTG objects. At least one NWSD object must be created in the host for each client, though more are supported.

The NWSSTG objects are the virtual disks provided to the client IBM i partition. They are created from available physical storage in the host partition. In the client, they are recognized and managed as standard DDxx disk devices (with a different type and model). The following screenshot shows several storage spaces for a client partition in an IBM i 6.1 host partition:

The next screenshot shows several storage spaces in an IBM i 6.1 client partition:

Storage spaces for an IBM i client partition do not have to match physical disk sizes; they can be created from 160 MB to 1 TB in size, as long as there is available storage in the host. The 160 MB minimum size is a requirement from the storage management Licensed Internal Code (LIC) on the client partition. For an IBM i client partition, up to 16 NWSSTGs can be linked to a single NWSD, and therefore, to a single virtual SCSI connection. Up to 32 outstanding I/O operations from the client to each storage space are supported for IBM i clients. Storage spaces can be created in any existing Auxiliary Storage Pool (ASP) on the host, including Independent ASPs. Through the use of NWSSTGs, any physical storage supported in the IBM i host partition on a

POWER6-based system can be virtualized to a client partition.

3.3. Optical virtualization

Any optical drive supported in the host IBM i LPAR can be virtualized to an IBM i client LPAR. An existing virtual SCSI connection can be used, or a new connection can be created explicitly for optical I/O traffic. By default, if a virtual SCSI connection exists between host and client, all physical OPTxx optical drives in the host will be available to the client, where they will also be recognized as OPTxx devices. The NWSD parameter Restricted device resources can be used to specify which optical devices in the host a client partition cannot access.

A virtualized optical drive in the host partition can be used for a D-mode Initial Program Load (IPL) and install of the client partition, as well as for installing Program Temporary Fixes (PTFs) or applications later. If the optical drive is writeable, the client partition will be able to write to the physical media in the drive.

3.4. Network virtualization

Virtualizing a network adapter and using a virtual LAN (VLAN) for partition-to-partition communication within a system are existing IBM i capabilities. In order for a client to use a host’s physical network adapter, a virtual Ethernet adapter must be created in the HMC in both partitions. To be on the same VLAN, the two virtual Ethernet adapters must have the same Port Virtual LAN ID (PVID). This type of adapter is recognized by IBM i as a communications port (CMNxx) with a different type. In the host partition, the virtual Ethernet adapter is then associated with the physical network adapter via a routing configuration – either Proxy ARP or Network Address Translation (NAT). This allows the client partition to send network packets via the VLAN and through the physical adapter to the outside LAN. The physical adapter can be any network adapter supported by IBM i 6.1, including Integrated Virtual Ethernet (IVE) ports, also known as Host Ethernet Adapter (HEA) ports.

4. Prerequisites for implementing IBM i hosted LPARs

4.1. Storage planning

Because virtual disks for the IBM i client LPAR are NWSSTG objects in the host LPAR, the main prerequisite to installing a new client LPAR is having sufficient capacity in the host to create those objects. Note that the host partition is not capable of detecting what percent of the virtual storage is used in the client. For example, if a 500-GB storage space is created, it will occupy that amount of physical storage in the host IBM i LPAR, even if the disk capacity is only 50% utilized in the client LPAR.

It is recommended to closely match the total size of the storage spaces for each client partition to its initial disk requirements. As the storage needs of the client partition grow, additional storage spaces can be dynamically created and linked to it on the host partition. On the client, the new virtual disk will automatically be recognized as a non-configured drive and can be added to any existing ASP. The only restriction to consider in this case is the maximum number of storage spaces allowed per virtual SCSI connection for an IBM i client partition, which is 16. If more than 16 NWSSTGs are needed for a client LPAR, additional virtual SCSI connections can be created dynamically in the HMC.

4.2. Performance

As described in section 2.2, disk I/O operations in an IBM i virtual client partition result in I/O requests to the physical disk adapter(s) and drives assigned to the host partition. Therefore, the

best way to ensure good disk performance in the client LPAR is to create a well-performing disk configuration in the host LPAR. Because the host partition is a standard IBM i partition, all the

recommendations in the Performance Capabilities Reference manual (http://www.ibm.com/systems/i/solutions/perfmgmt/resource.html manual’s suggestions for maximizing IBM i disk performance for the type of physical storage used in the host, whether it is integrated disk or SAN.

Note that if only the System ASP exists on the host partition, NWSSTG objects are created on the same physical disk units as all other objects. If the host partition is running production applications in addition to providing virtual storage to client partitions, there will be disk I/O contention as both client partitions and IBM i workloads in the host send I/O requests to those disk units. To minimize disk I/O contention, create storage space objects in a separate ASP on the host (Independent ASPs are supported). Performance on the client(s) would then depend on the disk adapter and disk configuration used for that ASP. If the host partition is providing virtual storage to more than one client partition, consider using separate ASPs for the storag e space objects for each client. This recommendation should be weighed against the concern of ending up with too few physical disk arms in each ASP to provide good performance.

Disk contention from IBM i workloads in the host LPAR and virtual client LPARs can be eliminated if a separate IBM i LPAR is used just for hosting client LPARs. An additional benefit of this configuration is the fact that an application or OS problem stemming from a different workload on the host cannot negatively affect client partitions. These benefits should be weighed against:

• The license cost associated with a separate IBM i partition

• The maintenance time required for another partition, such as applying Program

Temporary Fixes (PTFs)

• The ability to create well-performing physical disk configurations in both partitions that meet the requirements of their workloads

If the host partition runs a heavy-I/O workload and the client partitions also have high disk response requirements, it is strongly recommended to consider using a separate hosting partition, unless separate ASPs on the host are used for storage space objects. If the host partition’s workload is light to moderate with respect to disk requirements and the client partitions are used mostly for development, test or quality assurance (QA), it is acceptable to use one IBM i partition for both tasks.

4.3. Dual hosting

An IBM i client partition has a dependency on its host: if the host partition fails, IBM i on the client will lose contact with its disk units. The virtual disks would also become unavailable if the host partition is brought down to restricted state or shut down for scheduled maintenance or to apply

) will apply to it. Use the

PTFs. To remove this dependency, two host partitions can be used to simultaneously provide virtual storage to one or more client partitions.

The configuration for two hosts for the same client partition uses the same conc epts as that for a single host described in section 2.2. In addition, a second virtual SCSI client adapter exists in the client LPAR, connected to a virtual SCSI server adapter in the second host LPAR. The IBM i configuration of the second host mimics that of the first host, with the same number of NWSD and NWSSTG objects, and NWSSG objects of the same size. As a result, the client partition recognizes a second set of virtual disks of the same number and size. To achie v e redundancy, adapter-level mirroring is used between the two sets of storage spaces from the two hosts. Thus, if a host partition fails or is taken down for maintenance, mirroring will be suspended, but the client partition will continue to operate. When the inactive host is either recovered or restarted, mirroring can be resumed.

5. Implementing IBM i client LPARs with an IBM i host Installing IBM i in a client LPAR with an IBM i host consists of two main phases:

• Creating the virtual SCSI configuration in the HMC

• Creating the NWSSTG and NWSD objects in the IBM i host partition, and activating the

new client partition

The implementation steps are described in detail in the topic Creating an IBM i logical partition

that uses IBM i virtual I/O resources using the HMC in the Power Systems Logical Partitioning Guide: http://publib.boulder.ibm.com/infocenter/systems/scope/hw/topic/iphat/iphat.pdf

sufficient available capacity is required in the IBM i host partition to create the storage space objects. When following the detailed implementation instructions, keep in mind the performance recommendations in section 4.2 of this document.

6. Post-install tasks and considerations

6.1. Configure IBM i networking

Once the IBM i client partition is installed and running, the first system management step is to configure networking. There are three types of network adapters that can be assigned to an IBM i client partition:

• A standard physical network adapter in a PCI slot

• A logical port on a Host Ethernet Adapter (HEA)

• A virtual Ethernet adapter

Note that both physical and virtual I/O resources can be assigned to an IBM i virtual client partition. If a physical network adapter was not assigned to the IBM i client partition when it was first created, see the topic Managing physical I/O devices and slots dynamically using the HMC in the Power Systems Logical Partitioning Guide (http://publib.boulder.ibm.com/infocenter/systems/scope/hw/topic/iphat/iphat.pdf available adapter.

An IBM i client partition can also use the new HEA capability of POWER6 processor-based servers. To assign a logical port (LHEA) on an HEA to an IBM i client partition, see the topic Creating a Logical Host Ethernet Adapter for a running logical partition using the HMC in

. Note that

) to assign an

the Power Systems Logical Partitioning Guide: http://publib.boulder.ibm.com/infocenter/systems/scope/hw/topic/iphat/iphat.pdf.

Lastly, a virtual Ethernet adapter can also provide network connectivity to an IBM i client partition. To create one, consult the topic Configuring a virtual Ethernet adapter using the HMC in the Power Systems Logical Partitioning Guide: http://publib.boulder.ibm.com/infocenter/systems/scope/hw/topic/iphat/iphat.pdf

In all three cases, the assigned network adapter will be recognized as a communication s port (CMNxx) in IBM i. The type of communications port will depend on the network adapter: for example, 5706 for a Gigabit Ethernet adapter, 5623 for an LHEA and 268C for a virtual Ethernet adapter. In the case of a standard PCI network adapter or an LHEA, networking can be configured following the process described in the IBM i networking topic in the Information Center: http://publib.boulder.ibm.com/infocenter/systems/scope/i5os/topic/rzajy/rzajyoverview.htm

If the IBM i client partition is using a virtual Ethernet adapter for networking, additional configuration on the IBM i host is required. The virtual Ethernet adapter allows the client partition to communicate only with other partitions whose virtual Ethernet adapters have the same Port Virtual LAN ID (PVID); in other words, partitions on the same virtual LAN within the system. A routing configuration can be created in the IBM i host partition to allow forwarding of network packets from the outside LAN to the client partition on the virtual LAN. That type of virtual network configuration has been used successfully for several years to provide networking to Linux client partitions with an IBM i host. The two methods for routing traffic from the physical LAN to a client partition on a virtual LAN are Proxy ARP and Network Address Translation (NAT). To configure Proxy ARP or NAT in the IBM i host partition, follow the instructions in section 5.2 of the Redbook Implementing POWER Linux on IBM System i Platform (http://www.redbooks.ibm.com/redbooks/pdfs/sg246388.pdf

6.2. How to perform IBM i operator panel functions Operator panel functions in an IBM i client partitions are performed in the HMC:

• Sign onto the HMC with a profile with sufficient authority to manage the IBM i client partition

• Select the partition

• Use the open-in-context arrow to select Serviceability Æ Control Panel Functions, then

the desired function.

6.3. How to display the IBM i partition System Reference Code (SRC) history

• Sign onto the HMC with a profile with sufficient authority to manage the IBM i client partition

• Select the partition

• Use the open-in-context arrow to select Serviceability Æ Reference Code History

• To display words 2 through 9 of a reference code, click the radio button for that code.

6.4. Client IBM i LPARs considerations and limitations Consult the topic Considerations and limitations for i5/OS client partitions on systems

managed by the Integrated Virtualization Manager (IVM) in the Information Center:

http://publib.boulder.ibm.com/infocenter/systems/scope/i5os/topic/rzahc/rzahcbladei5limits.htm

While in this case the IBM i client partition is not being managed by IVM, it does virtual I/O resources and the limitations outlined in the topic above apply to it.

6.5. Configuring Electronic Customer Support (ECS) over LAN

A supported WAN adapter can be assigned to the IBM i client partition for ECS. Alternatively, ECS over LAN can be configured. Consult the topic Setting up a connection to IBM in the Information Center: http://publib.boulder.ibm.com/infocenter/systems/scope/i5os/topic/rzaji/rzaji_setup.htm

6.6. Copying storage spaces

Because an IBM i client partition is installed into one or more storage space objects in the IBM i host partition, new client partitions can be deployed rapidly by copying the storage space(s). Note that each IBM i partition, client or host, must have valid OS and Licensed Product Program licenses for the number of processors it uses.

To copy one or more storage spaces that contain an installed IBM i client partition, first shut down the partition during an available period of downtime. Next, log into the host IBM i partition with a security officer-level profile and perform the following steps:

• Enter WRKNWSSTG

• Enter 3 next to the storage space you are going to copy, then press Enter

• Enter a name of up to 10 characters for the new storage space

• The size of the original storage space will be entered automatically. The new storage

space can be as large or larger (up to 1 TB), but not smaller

• Enter the correct ASP ID. The ASP where the original storage space exists is the default

• Optionally, enter a text description

• Press Enter

To deploy the new client partition, follow the instructions in section 5 to create the necessary virtual SCSI configuration in the HMC and the NWSD object in the host IBM i partition.

6.7. Backups

As mentioned above, an IBM i client partition with an IBM i host can use a mix of virtual and physical I/O resources. Therefore, the simplest backup and restore approach is to assign an available tape adapter on the system to it and treat it as a standard IBM i partition. The tape adapter can be any adapter supported by IBM i on IBM Power servers and can be shared with other partitions. To assign an available tape adapter to the IBM i client partition, consult the topic

Managing physical I/O devices and slots dynamically using the HMC in the Logical Partitioning Guide

(http://publib.boulder.ibm.com/infocenter/systems/scope/hw/topic/iphat/iphat.pdf Once a tape adapter connected to a tape drive or library is available to the client partition, use the

Backup and Recovery topic in the Information Center to manage backups: http://publib.boulder.ibm.com/infocenter/systems/scope/i5os/index.jsp?topic=/rzahg/rzahgbackup. htm&tocNode=int_215989.

The IBM i host partition can also be used for system-level backups of the client partition. See the topic Saving IBM i server objects in IBM i in the Logical Partitioning Guide (http://publib.boulder.ibm.com/infocenter/systems/scope/hw/topic/iphat/iphat.pdf

7. IBM i using open storage supported configurations

There are three general methods by which IBM i connects to open storage:

• Directly through the SAN fabric without VIOS

• As a client of VIOS, with VIOS connecting to a storage subsystem through the SAN fabric

• As a client of VIOS, with VIOS using storage from a SAN Volume Controller (SVC). SVC

in turn connects to one or more storage subsystems through the SAN fabric

The first set of support tables in the hardware and software sections below applies to the first connection method above (IBM i direct attachment), while the second set of tables in both sections applies to the second method (IBM i Æ VIOS Æ storage subsystem). Furthermore, the

support statements in the second set of tables below apply to the end-to-end solution of IBM i using open storage as a client of VIOS. This document will not attempt to list the full

device support of VIOS, nor of any other clients of VIOS, such as AIX and Linux. For the general VIOS support statements, including other clients, see the VIOS Datasheet at: http://www14.software.ibm.com/webapp/set2/sas/f/vios/documentation/datasheet.html

When the second connection method to open storage is used (IBM i Æ VIOS Æ SVC Æ storage subsystem), the support statement for IBM i follows that of SVC. For a list of environments supported by SVC, see the data sheet on the SVC overview Web site: http://www.ibm.com/systems/storage/software/virtualization/svc directly to SVC; it must do so as a client of VIOS.

IBM i as a client of VIOS is also supported on POWER6 or later processor-based blade servers. IBM i cannot attach directly to open storage when running on Power blades. For the full supported configurations statement for the IBM i on Power blade solution, see the Supported Environments document at: http://www.ibm.com/systems/power/hardware/blades/ibmi.html

7.1. Hardware

Hardware type Supported by

IBM i for direct

attachment

IBM Power servers Yes Includes IBM Power 520 Express, IBM

Power 550 Express, IBM Power 560

Express, IBM Power 570 and IBM Power 595 IBM Power 575 No POWER5-based systems or earlier DS5100 using Fibre Channel or SATA drives

DS5300 using Fibre Channel or SATA drives

EXP810 expansion unit with Fibre Channel or SATA drives

EXP5000 expansion unit with Fibre Channel or SATA drives

Yes It is strongly recommended that SATA

drives are used only for test or archival IBM i

applications for performance reasons

Yes It is strongly recommended that SATA

drives are used only for test or archival IBM i

applications for performance reasons

Yes It is strongly recommended that SATA

drives are used only for test or archival IBM i

applications for performance reasons

Yes It is strongly recommended that SATA

drives are used only for test or archival IBM i

applications for performance reasons

. Note that IBM i cannot connect

Notes

EXP5060 expansion unit with Fibre Channel or SATA drives

DS6800 Yes DS8100 using Fibre Channel or FATA drives

DS8300 using Fibre Channel or FATA drives

DS8700 using Fibre Channel or FATA drives

Hardware type Supported by

IBM Power servers Yes Includes IBM Power 520 Express, IBM

IBM Power 575 No POWER5-based systems or earlier DS3950 using Fibre Channel or SATA drives

DS4800 using Fibre Channel drives

DS4800 using SATA drives Yes It is strongly recommended that SATA

DS4700 using Fibre Channel drives

DS4700 using SATA drives Yes It is strongly recommended that SATA

DS3400 using SAS or SATA drives

DS4200 and earlier DS4000 and FAStT models EXP810 expansion unit attached to DS4800 or DS47000 EXP710 expansion unit attached to DS4800 or DS4700 SAN Volume Controller (SVC) Yes SVC is not supported for direct connection to

SVC Entry Edition (SVC EE) Yes SVC EE is not supported for direct

Yes It is strongly recommended that SATA

drives are used only for test or archival IBM i

applications for performance reasons

Yes It is strongly recommended that FATA

drives are used only for test or archival IBM i

applications for performance reasons

Yes It is strongly recommended that FATA

drives are used only for test or archival IBM i

applications for performance reasons

Yes It is strongly recommended that FATA

drives are used only for test or archival IBM i

applications for performance reasons

Notes

IBM i as a

client of VIOS

Power 550 Express, IBM Power 560

Express, IBM Power 570 and IBM Power 595

Yes It is strongly recommended that SATA

drives are used only for test or archival IBM i

applications for performance reasons

Yes Supported by IBM i as a client of VIOS on

both IBM Power servers and IBM Power

blade servers

drives are used only for test or archival IBM i

applications for performance reasons

Yes Supported by IBM i as a client of VIOS on

both IBM Power servers and IBM Power

blade servers

drives are used only for test or archival IBM i

applications for performance reasons

Yes It is strongly recommended that SATA

drives are used only for test or archival IBM i

applications for performance reasons

Yes See supported drives comments above

IBM i

connection to IBM i

Storage subsystems connected to SVC DS5020 using Fibre Channel or SATA drives

DS5100 using Fibre Channel or SATA drives

DS5300 using Fibre Channel or SATA drives

DS8100 using Fibre Channel or FATA drives

DS8300 using Fibre Channel or FATA drives

DS8700 using Fibre Channel or FATA drives

XIV Storage System Yes Fibre Channel adapters Yes Must be supported by VIOS*

*For a list of Fibre Channel adapters supported by VIOS, see the VIOS Datasheet at: http://www14.software.ibm.com/webapp/set2/sas/f/vios/documentation/datasheet.html

7.2. Software and firmware

Software or firmware type Supported by

IBM i 6.1 with LIC 6.1.1 or later Yes IBM i 6.1 without LIC 6.1.1 No IBM i 5.4 or earlier No DS5000 Copy Services Yes See Section 16.5 of this document for details IBM i host kit for DS5000 Yes This license is required in order to attach

Controller firmware

07.60.26.00 or later DS Storage Manager

10.60.x5.16 or later

Software or firmware type Supported by

IBM i 6.1 or later Yes IBM i 5.4 or earlier No PowerVM Standard Edition or higher VIOS 1.5 with Fix Pack 10.1 or Yes This is the minimum level of VIOS required

Yes The supported list for IBM i as a client of

VIOS follows that for SVC

Yes It is strongly recommended that SATA

drives are used only for test or archival IBM i

applications for performance reasons

Yes It is strongly recommended that SATA

drives are used only for test or archival IBM i

applications for performance reasons

Yes It is strongly recommended that SATA

drives are used only for test or archival IBM i

applications for performance reasons

Yes It is strongly recommended that FATA

drives are used only for test or archival IBM i

applications for performance reasons

Yes It is strongly recommended that FATA

drives are used only for test or archival IBM i

applications for performance reasons

Yes It is strongly recommended that FATA

drives are used only for test or archival IBM i

applications for performance reasons

Notes

IBM i for direct

attachment

DS5000 LUNs directly to IBM i

Yes Yes

Notes

IBM i as a

client of VIOS

Yes Required for VIOS media and activation code

later IBM Power server system firmware 320_040_031 or later HMC firmware HMC V7 R3.2.0 or later DS4000 controller firmware

6.60.08 or later DS4000 Storage Manager 9.x

or later DS4000 Copy Services Yes See Section 12 of this document for details DS5000 Copy Services Yes See Section 12 of this document for details AIX/VIOS host kit license for DS4000

SVC code level 4.3.1 Yes This is the minimum SVC code level required SVC Copy Services Yes See Section 14 of this document for details XIV firmware 10.0.1b Yes This is the minimum XIV firmware level

XIV Copy Services Yes See Section 15 of this document for details

8. IBM i using open storage through VIOS concepts

The capability to use open storage through VIOS extends the IBM i storage portfolio to include 512-byte-per-sector storage subsystems. The existing IBM i storage portfolio includes integrated SCSI, SAS or Solid State disk, as well as Fibre Channel-attached storage subsystems that support 520 bytes per sector, such as the IBM DS8000 product line. IBM i cannot currently attach directly to 512-B open storage, because its Fibre Channel adapter code requires 520 bytes per sector. Therefore, logical units (LUNs) created on open storage are physically connected through the Fibre Channel fabric to VIOS. Once recognized by VIOS, the LUNs are virtualized to one or more IBM i virtual client partitions. The following diagram illustrates both the physical and virtualization components of the solution:

Yes This is the minimum system firmware level

required

Yes This is the minimum HMC firmware level

required

Yes This is the minimum DS4000 controller

firmware level required. Controller

firmware 7.10.23 or later is supported.

Yes Storage Manager 10.10.x or later is

supported

Yes This license is required in order to attach

DS4000 LUNs to VIOS, and therefore

virtualize them to IBM i

required

VIOS has been used successfully for several years to virtualize storage, optical and networking resources to AIX and Linux client partitions. Now IBM i joins this virtualization environment, gaining the ability to use open storage. From a VIOS perspective, IBM i is another client partition; the host-side configuration steps are the same as for AIX and Linux clients. While code changes were made in VIOS to accommodate IBM i client partitions, there is no special version of VIOS in use for IBM i. If you have existing skills in attaching open storage to VIOS and virtualizing I/O resources to client partitions, they will continue to prove useful when creating a configuration for an IBM i client partition.

The hardware and virtualization components for attaching open storage to IBM i illustrated in the diagram above also apply to using DS5000, XIV and other subsystems supported for this solution, as listed in section 7.

Two management interfaces are available for virtualizing I/O resources through VIOS to IBM i: the Hardware Management Console (HMC) and the Integrated Virtualization Manager (IVM). Both provide logical partitioning and virtualization management functions, however, their support on Power servers and Power blades is different:

• The HMC is supported with all Power servers, from Power 520 Express to Power 595, but not supported with Power blades, such as IBM BladeCenter JS12 or JS23

• IVM is supported with all Power blades, from JS12 to JS43, and only with Power 520 and Power 550

There is another significant difference between HMC and IVM when managing I/O virtualization for IBM i: when the Power server is IVM-managed, IBM i partitions must be purely virtual. They

are not allowed to own any physical I/O adapters. Therefore, when managing Power servers, the HMC is used most of the time, with IVM being used occasionally only on Power 520 and 550. When managing Power blades, IVM must always be used.

8.1. Virtual SCSI and Ethernet adapters

VIOS providing storage to an IBM i client LPAR uses an existing function of the system firmware, or Power Hypervisor: the capability to create virtual SCSI and Ethernet adapters in a partition. Virtual adapters are created for each LPAR in the HMC. Virtual SCSI adapters are used for storage and optical virtualization; virtual Ethernet adapters are used for network virtualization.

Note that using virtual I/O resources from VIOS does not preclude an IBM i client partition from owning physical hardware when the server is HMC-managed. A mix of virtual and physical hardware in the same partition is supported for IBM i in this environment, by assigning both types of adapters to the partition in the HMC.

As mentioned above, using IVM for LPAR and virtualization management mandates using only virtual resources for IBM i partitions. However, IVM provides the benefit of automatically creating Virtual SCSI server (in VIOS) and client adapters (in IBM i) as virtual disk, optical and tape resources are assigned to IBM i. For more information on LPAR management with IVM, consult the IVM topic in the PowerVM Editions Guide, available at: http://publib.boulder.ibm.com/infocenter/systems/scope/hw/index.jsp?topic=/arecu/arecukickoff.ht m.

Note that IVM will automatically create only the first pair of Virtual SCSI server and client adapters when disk and optical resources are virtualized to IBM i. Additional adapter pairs must be created by using the IVM command line, available via a Telnet or SSH session to VIOS. Consult the VIOS topic in the PowerVM Editions Guide for more information on using the IVM/VIOS command line. IVM will automatically create a new Virtual SCSI adapter pair for each tape resource assigned to IBM i.

8.2. Storage virtualization

8.2.1. HMC and IVM configuration

For VIOS to virtualize LUNs created on open storage to an IBM i client partition, configuration takes place both in the HMC (or IVM) and in VIOS. As mentioned in the previous section, no explicit Virtual SCSI configuration is necessary in IVM the majority of the time.

In the HMC, the minimum required configuration is:

• One virtual SCSI server adapter in the host partition

• One virtual SCSI client adapter in the client partition

This virtual SCSI adapter pair allows the client partition to send read and write I/O operations to the host partition. More than one virtual SCSI pair can exist for the same client partition in this environment. To minimize performance overhead in VIOS, the virtual SCSI connection is used to send I/O requests, but not for the actual transfer of data. Using the capability of the Power Hypervisor for Logical Remote Direct Memory Access (LRDMA), data are transferred directly from the Fibre Channel adapter in VIOS to a buffer in memory of the IBM i client partition.

In an IBM i client partition, a virtual SCSI client adapter is recognized as a type 290A DCxx storage controller device. The screenshot below depicts the virtual SCSI client adapter, as well as several open storage LUNs and an optical drive virtualized by VIOS:

In VIOS, a virtual SCSI server adapter is recognized as a vhostX device:

8.2.2. VIOS configuration and HMC enhancements

No additional configuration in the IBM i client partition is necessary. In VIOS, however, a new object must be created for each open storage LUN that will be virtualized to IBM i: a virtual target SCSI device, or vtscsiX. A vtscsiX device makes a storage object in VIOS available to I B M i as a standard DDxxx disk unit. There are three types of VIOS storage objects that can be virtualized to IBM i:

• Physical disk units or volumes (hdiskX), which are open storage LUNs in this case

• Logical volumes (hdX and other)

• Files in a directory

For both simplicity and performance reasons, it is recommended to virtualize open storage LUNs to IBM i directly as physical devices (hdiskX), and not through the use of logical

volumes or files. (See section 9.2 for a detailed performance discussion.) A vtscsiX device lin ks a LUN available in VIOS (hdiskX) to a specific virtual SCSI adapter (vhostX). In turn, the virtual SCSI adapter in VIOS is already connected to a client SCSI adapter in the IBM i client partition. Thus, the hdiskX LUN is made available to IBM i through a vtscsiX device. What IBM i storage management recognizes as a DDxxx disk unit is not the open storage LUN itself, but the corresponding vtscsiX device. The vtscsiX device correctly reports the parameters of the LUN, such as size, to the virtual storage code in IBM i, which in turn passes them on to storage management.

Multiple vtscsiX devices, corresponding to multiple open storage LUNs, can be linked to a single vhostX virtual SCSI server adapter and made available to IBM i. Up to 16 LUNs can be virtualized to IBM i through a single virtual SCSI connection. Each LUN typically uses multiple physical disk arms in the open storage subsystem. If more than 16 LUNs are required in an IBM i

client partition, an additional pair of virtual SCSI server (VIOS) and client (IBM i) adapters must be created in the HMC or on the IVM/VIOS command line. Additional LUNs available in VIOS can then be linked to the new vhostX device through vtscsiX devices, making them available to IBM i.

Prior to May 2009, creating vtscsiX devices and thus virtualizing open storage LUNs to IBM i was necessarily a task performed only on the IVM/VIOS command line when the server was HMCmanaged. When the server is IVM-managed, assignment of virtual resources is performed using the IVM browser-based interface. However, the HMC interface has been significantly enhanced, making the use of the IVM/VIOS command line rare. Once available in VIOS, open storage LUNs and optical devices can be assigned to IBM i using the following steps in th e HMC:

• Select the correct managed server

• In the menu below, expand Configuration, then Virtual Resources

• Click Virtual Storage Management

• Select the correct VIOS in the drop-down menu and click Query VIOS

• Click the Physical Volumes (or Optical Devices for optical resources)

• Select the correct LUN(s) or optical device(s) and click Modify assignment… to assign

to IBM i

If the QAUTOCFG system value in IBM i is set to “1” (which is the default), the new virtual resources will become available in IBM i immediately. No steps are required on the IVM/VIOS command line.

8.2.3. Adding devices to VIOS

VIOS does not autoscan for new devices added after the initial boot is complete. When a new device – such as a LUN on a storage subsystem – is added to VIOS, prompt VIOS to scan for new devices by connecting with Telnet or SSH and using the following command:

• cfgdev

Once it has detected the device, VIOS will autoconfigure it and make it available for use. It can then be virtualized to the IBM i client partition using the HMC, as described in the previous section. Note that if the SAN configuration is performed before the VIOS partition boots, this step is not necessary, as VIOS will recognize all available devices at boot time.

8.3. Optical virtualization

8.3.1. Physical optical devices

CD and DVD drives supported by VIOS can be virtualized to IBM i directly. Optical drives are recognized in VIOS as cdX devices. A cdX device is linked to a vhostX virtual SCSI server adapter and made available to IBM i through a virtual target optical device, vtoptX. The same optical drive can be made virtualized to multiple IBM i client partitions by creating separate vtoptX devices linked to different vhostX adapters for the same cdX device. Only one IBM i client partition can use the physical optical drive at a time. If the physical optical drive is writeable, IBM i will be able to write to it. Similar to LUNs, optical devices can be virtualized to IBM i using the enhanced functions of the HMC; it is not necessary to use the IVM/VIOS command line to create the vtoptX devices.

8.3.2. VIOS media repository

VIOS provides a capability similar to that of an image catalog (IMGCLG) in IBM i: a repository of media images on disk. Unlike IMGCLGs in IBM i, a single media repository may exist per VIOS. The media repository allows file-backed virtual optical volumes to be made available to the IBM i client partition through a separate vtoptX device. One immediate benefit to IBM i is the ability to import the IBM i install media into VIOS as ISO images, then install the client partition from the ISO images instead of switching real media in a physical DVD drive.

8.4. Network virtualization

Virtualizing a network adapter and using a virtual LAN (VLAN) for partition-to-partition communication within a system are existing Power server capabilities. In order for an IBM i client to use a physical network adapter in VIOS, a virtual Ethernet adapter must be created in both partitions in the HMC. To be on the same VLAN, the two virtual Ethernet adapters must have the same Port Virtual LAN ID (PVID).

A virtual Ethernet adapter is recognized by IBM i as a communications port (CMNxx) of type 268C:

In VIOS, the same Ethernet type of device, entX, is used for logical Host Ethernet ports, physical and virtual Ethernet adapters:

VIOS provides virtual networking to client partitions, including IBM i, by bridging a physical Ethernet adapter and one or more virtual Ethernet adapters. The virtualization object that provides this Ethernet bridge is called a Shared Ethernet Adapter (SEA). The SEA forwards network packets from any client partitions on a VLAN to the physical LAN through the physical Ethernet adapter. Because the SEA creates a Layer-2 bridge, the original MAC address of the virtual Ethernet adapter in IBM i is used on the physical LAN. The CMNxx communications port that represents the virtual Ethernet adapter in IBM i is configured with an externally routable IP address and a standard network configuration is used. The physical adapter bridge by the SEA can be any network adapter supported by VIOS, including Integrated Virtual Ethernet (IVE) ports, also known as Host Ethernet Adapter (HEA) ports.

The enhanced virtualization management functions of the HMC also allow for net work virtualization without using the IVM/VIOS command line. Follow these steps in the HMC:

• Select the correct managed server

• In the menu below, expand Configuration, then Virtual Resources

• Click Virtual Network Management

• Use the drop-down menu to create or modify the settings of a VLAN (referred to as a

VSwitch), including which SEA it uses to access the external network

9. Prerequisites for attaching open storage to IBM i through VIOS

9.1. Storage planning

The first storage planning consideration is having enough available capacity in the open storage subsystem to create the AIX/VIOS LUNs that will be virtualized to IBM i by VIOS. As mentioned in the supported hardware section (7.1), it is strongly recommended that only Fibre Channel or SAS physical drives are used to create LUNs for IBM i as a client of VIOS. The reason is the performance and reliability requirements of IBM i production workloads. For non-I/O-intensive workloads or nearline storage, SATA or FATA drives may also be used. This recommendation is not meant to preclude the use of SATA or FATA drives for other clients of VIOS or other host servers; it applies only to production IBM i workloads.

If the storage subsystem requires a host kit for AIX/VIOS to be installed before attaching LUNs to these hosts (as is the case with DS5000 and DS4000), that host kit is also required when virtualizing LUNs to IBM i. Because the LUNs are virtualized by VIOS, they do not have to match IBM i integrated disk sizes. The technical minimum for any disk unit in IBM i is 160 MB and the maximum is 2 TB, as measured in VIOS. Actual LUN size will be based on the capacity and performance requirements of each IBM i virtual client partition.

9.2. Performance

When creating an open storage LUN configuration for IBM i as a client of VIOS, it is crucial to plan for both capacity and performance. It may seem that because LUNs are virtualized for IBM i by VIOS instead of being directly connected, the virtualization layer will necessarily add a significant performance overhead. However, internal IBM performance tests clearly show that the VIOS layer adds a negligible amount of overhead to each I/O operation. Instead, our tests demonstrate that when IBM i uses open storage LUNs virtualized by VIOS, performance is

almost entirely determined by the physical and logical configuration of the storage subsystem.

The IBM Rochester, MN, performance team has run a significant number of tests with IBM i as a client of VIOS using open storage. The resulting recommendations on configuring both the open storage and VIOS are available in the latest Performance Capabilities Reference manual (PCRM) at http://www.ibm.com/systems/i/solutions/perfmgmt/resource.html virtualized storage for IBM i. In most cases, an existing IBM i partition using physical storage will be migrated to open storage LUNs virtualized by VIOS. The recommended approach here is to start with the partition’s original physical disk configuration; then create a similar setup with the physical drives in the open storage subsystem on which LUNs are created, while following the suggestions in the PCRM sections above.

The commonly used SAN disk sizing tool Disk Magic can also be used to model the projecte d IBM i performance of different physical and logical drive configurations on supported subsystems. Work with IBM Techline or your IBM Business Partner for a Disk Magic analysis. The latest

. Chapter 14.6 focuses on

version of Disk Magic includes support for multiple open storage subsystems and IBM i as a virtual client of VIOS.

9.3. Dual hosting and multi-path I/O (MPIO)

An IBM i client partition in this environment has a dependency on VIOS: if the VIOS partition fails, IBM i on the client will lose contact with the virtualized open storage LUNs. The LUNs would also become unavailable if VIOS is brought down for scheduled maintenance or a release upgrade. To remove this dependency, two or more VIOS partitions can be used to simultaneously provide virtual storage to one or more IBM i client partitions.

9.3.1. Dual VIOS LPARs with IBM i mirroring

Prior to the availability of redundant VIOS LPARs with clients-side MPIO for IBM i, the only method to achieve VIOS redundancy was to use mirroring within IBM i. This configuration uses the same concepts as that for a single VIOS described in section 8.2. In addition, at least one additional virtual SCSI client adapter exists in the client LPAR, connected to a virtual SCSI server adapter in the second VIOS on the same Power server. A second set of LUNs of the same number and size is created on the same or a different open storage subsystem, a nd connected to the second VIOS. The host-side configuration of the second VIOS mimics that of the first host, with the same number of LUNs (hdisks), vtscsiX and vhostX devices. As a result, the client partition recognizes a second set of virtual disks of the same number and size. To achieve redundancy, adapter-level mirroring in IBM i is used between the two sets of virtualized LUNs from the two hosts. Thus, if a VIOS partition fails or is taken down for maintenance, mirroring will be suspended, but the IBM i client will continue to operate. When the inactive VIOS is either recovered or restarted, mirroring can be resumed in IBM i.

9.3.2. Path redundancy to a single set of LUNs

Note that the dual-VIOS solution above provides a level of redundancy by attaching two separate sets of open storage LUNs to the same IBM i client through separate VIOS partitions. It is not an

MPIO solution that provides redundant paths to a single set of LUNs. There are two MPIO scenarios possible with VIOS that remove the requirement for two sets of LUNs:

• A single VIOS partition using two Fibre Channel adapters to connect to the same set of

LUNs

• Two VIOS partitions providing redundant paths to the same set of LUNs on a single open

storage subsystem

9.3.2.1. Path redundancy with a single VIOS

If a VIOS LPAR has two or more Fibre Channel adapters assigned and the correct host configuration is created in the open storage subsystem, VIOS will have redundant paths to the LUNs connected to the Fibre Channel adapters. VIOS includes a basic MPIO driver, which has been the default instead of the RDAC (Redundant Disk Array Controller) driver since November

2008. The MPIO driver can be used with any storage subsystem which VIOS supports and is included in a default install. In this case, configuration is required only on the storage subsystem in order to connect a single set of LUNs to both ports on a Fibre Channel adapter owned by VIOS.

This multi-path method can be configured in two ways: round-robin, or failover. For storage systems like XIV and DS8000 that support active-active multi-path connections, either method is allowed. Path failover will use one path and leave the other idle until the first path fails, while round-robin will always attempt to use both paths if available to maximize throughput. For

systems that only support active-passive connections, such as IBM’s DS3000, DS4000, and DS5000 subsystems, failover is the only method allowed.

The access method can be configured on a per-LUN basis from the VIOS command line via a Telnet or SSH session. To show the current multi-path algorithm:

• lsdev –dev hdiskX –attr algorithm

To change the multi-path algorithm:

• chdev –dev hdiskX –attr algorithm=round_robin, or

• chdev –dev hdiskX –attr algorithm=fail_over

These commands must be repeated for each hdisk.

9.3.2.2. Redundant VIOS LPARs with client-side MPIO

Beginning with IBM i 6.1 with Licensed Internal Code (LIC) 6.1.1, the IBM i Virtual SCSI (VSCSI) client driver supports MPIO through two or more VIOS partitions to a single set of LUNs (up to a maximum of 8 VIOS partitions). This multi-path configuration allows a VIOS partition to fail or be brought down for service without IBM i losing access to the disk volumes since the other VIOS partition(s) remain active. The following diagram illustrates the new capability for IBM i:

Note that as with dual VIOS LPARs and IBM i mirroring, a VSCSI connection from IBM i to both VIOS partitions is required. On the storage subsystem, it is recommended to create a host for each VIOS containing the World-wide Port Names (WWPNs) of all Fibre Channel adapters in that LPAR, and then create a host group comprising those VIOS hosts. A single set of LUNs for IBM i can then be created and mapped to the host group, giving both VIOS LPARs access to the LUNs. Configuring redundant VIOS LPARs with client-side MPIO does not preclude using MPIO within a VIOS for a higher level of resiliency, as described in the previous section.

For storage systems such as DS8000 and XIV, these connections are typically shared via the round-robin access as described above, so it is good practice to have two or more connections to each VIOS partition. For DS3000, DS4000 and DS5000 with dual controllers, a connection must be made to both controllers to allow an active and a failover path. When the volumes are created on these systems, the host OS type should be set to DEFAULT or AIX (not AIX ADT/AVT, or failover/failback oscillations could occur). For all storage systems, it is recommende d that the fabric configuration uses separate dedicated zones and FC cables for each connection.

In addition to the VSCSI and storage subsystem configuration, there are several VIOS settings that must be changed. These must be set via a Telnet or SSH session on both VIOS partitions. The first group of settings is applied to each of the fscsiX devices being used. To show the current fscsiX devices:

• lspath

Set the fast fail and dynamic tracking attributes on these devices:

• chdev –dev fscsiX –attr fc_err_recov=fast_fail –perm

• chdev –dev fscsiX –attr dyntrk=yes –perm

Next, restart VIOS. The pair of chdev commands must be repeated for each fscsiX device but only one restart is required after all fscsiX attributes have been set.

To verify the settings have been changed successfully:

• lsdev –dev fscsiX –attr fc_err_recov

• lsdev –dev fscsiX –attr dyntrk

Importantly, the SCSI reserve policy for each LUN (or hdisk) on both VIOS LPARs must be set to no_reserve to allow disk sharing. Some storage subsystems such as XIV default to no_reserv e

and do not require a change, while others such as DS4000 and DS5000 default to single_path. The change must be made prior to mapping the LUNs to IBM i and it does not require a restart of VIOS.

To show the current reserve policy settings:

• lsdev –dev hdiskX –attr reserve_policy

To set the no_reserve attribute if necessary:

• chdev –dev hdiskX –attr reserve_policy=no_reserve

Once the VSCSI client driver in IBM i detects a second path to the same set of LUNs through a different VIOS LPAR, the disk names change from DDxxx to DMPxxx. This is identical to multipath I/O to a directly attached DS8000 subsystem. If a path is lost, the disk names do not change back to DDxxx, but a path failure message is sent to the QSYSOPR message queue in IBM i.

Once the path is restored, an informational message is sent to the same queue. To monitor the status of the paths in IBM i:

• Start Services Tools and sign on

• Work with disk units

• Display disk configuration

• Display disk path status

Note that while the HMC allows a single VSCSI client adapter to be tagged as an IPL device, it is not required to change the tagged resource if that path is lost. As long as the alternate path through the second VIOS is active, the IBM i LPAR will be able to IPL. from the same load source LUN. The alternate path must exist prior to the loss of the original IPL path and prior to powering the IBM i LPAR off.

9.3.3 Subsystem Device Driver – Path Control Module (SDDPCM)

VIOS also supports the Subsystem Device Driver – Path Control Module (SDDPCM) for certain storage subsystems. Examples of supported subsystems include the SAN Volume Controller (SVC) and IBM DS8000. To find out whether a particular storage system supports SDD-PCM for VIOS, consult its interoperability matrix on the SDDPCM Web site: http://www.ibm.com/support/docview.wss?uid=ssg1S4000201. Note that there are separate support statements for AIX and VIOS. If SDDPCM is supported on your storage subsystem for VIOS, download and install the driver following the instructions in the Multi-path Subsystem Device Driver User's Guide at the same location.

10. Attaching open storage to IBM i through VIOS As described in section 8, IBM i joins the VIOS virtualization environment, allowing it to use open

storage. The setup process involves three main steps:

• Open storage configuration

• VIOS installation and configuration

• IBM i installation and configuration

The VIOS and open storage configuration steps are the same as for existing clients of VIOS, such as AIX and Linux. These steps are well documented and any existing skills in those areas would apply to IBM i as a client partition, as well. The only significant difference with IBM i is the specific open storage configuration requirements in order to achieve good performance, as referenced in section 9.2.

10.1. Open storage configuration All open storage subsystems follow similar general configuration steps:

• Perform physical disk and RAID configuration

• Create volumes (LUNs)

• Attach those volumes to a host (VIOS in this case)

Naturally, the detailed steps and graphical interfaces used vary by subsystem. Locate the correct documentation for your supported subsystem at http://www.redbooks.ibm.com follow the instructions for creating and attaching LUNs to an AIX host.

. Importantly,

10.2. VIOS installation and configuration

Consult the VIOS planning, installation and configuration topics in the PowerVM Editions Guide, available at: http://publib.boulder.ibm.com/infocenter/systems/scope/hw/index.jsp?topic=/arecu/arecukickoff.ht m.

As described in section 8.2.2, once VIOS is installed and physical I/O resources are assigned to it, those resources can be virtualized to IBM i using the HMC. It is not necessary to use the IVM/VIOS command line to assign virtual disk, optical and network resources to IBM i.

10.3. IBM i installation and configuration

The IBM i client partition configuration as a client of VIOS is the same as that for a client of an IBM i 6.1 host partition. Consult the topic Creating an IBM i logical partition that uses IBM i virtual I/O resources using the HMC in the Logical Partitioning Guide: http://publib.boulder.ibm.com/infocenter/systems/scope/hw/topic/iphat/iphat.pdf

10.4. HMC provisioning of open storage in VIOS

In May 2009, IBM enhanced the HMC’s capabilities to allow mapping of storage available in VIOS to client LPARs, such as IBM i. This significant improvement enables graphical provisioning of open storage for IBM i through VIOS and removes the requirement to use the VIOS command line. If the storage configuration is performed while VIOS is down or not yet installed, new storage will be discovered upon boot and the HMC can then be used to assign the correct LUNs to IBM i. If storage changes are made while VIOS and IBM i are running, it will still be necessary to Telnet to VIOS and run the cfgdev command to discover newly added storage prior to using the HMC for assigning it to IBM i.

For HMC provisioning of VIOS storage to work, the VIOS LPAR must have an active RMC (Resource Monitoring and Control) connection to the HMC. The same connection is required for performing Dynamic LPAR (DLPAR) changes to the VIOS LPAR and it generally involves successful TCP/IP communication between the LPAR and the HMC. If the HMC displays an RMC error while attempting VIOS storage provisioning, consult the DLPAR Checklist at: http://www.ibm.com/developerworks/systems/articles/DLPARchecklist.html

To provision open storage in VIOS to IBM i in the HMC, use the following steps:

• Sign onto the HMC as hscroot or another superadministrator-level userid

• Expand Systems Management

• Click Servers

• Select the correct managed system (server) by using the checkbox

• Use the launch-in-context arrow and select Configuration Æ Virtual Resources Æ

Virtual Storage Management

• Select the correct VIOS in the drop-down menu and click Query VIOS

• Click Physical Volumes

• Select the first correct LUN to be assigned to IBM i and click Modify assignment…

• Select the correct IBM i LPAR in the drop-down menu and click OK

• Repeat for all LUNs to be assigned to IBM i

Note that this sequence assumes the existence of the correct VSCSI server and client adapters in VIOS and IBM i, respectively. Once the assignment is performed, the new LUNs are immediately available in IBM i as non-configured drives.

10.5. End-to-end LUN device mapping

In October 2009, IBM enhanced both the HMC and VIOS to allow end-to-end device mapping for LUNs assigned to client LPARs, such as IBM i. The new function enables administrators to quickly identify which LUN reporting in VIOS (or, hdisk) is which DDxxx disk device in IBM i. This in turn makes it easier to troubleshoot disk-related problems and safer to change a virtualized disk configuration. In order to correctly perform the mapping, the HMC requires an active RMC connection to VIOS. Such a connection is not required for the IBM i client LPAR, because IBM i does not use RMC for DLPAR operations. However, if the client is an AIX LPAR, an active RMC connection to AIX is also required for a complete end-to-end mapping.

To perform end-to-end LUN device mapping, use the following steps:

• Sign onto the HMC as hscroot or another superadministrator-level userid

• Expand Systems Management

• Expand Servers

• Click the correct managed system (server)

• Select the correct VIOS by using the checkbox

• Use the launch-in-context arrow and select Hardware Information Æ Virtual I/O

Adapters Æ SCSI

11. Post-install tasks and considerations

11.1. Configure IBM i networking

• A standard physical network adapter in a PCI slot

• A logical port on a Host Ethernet Adapter (HEA)

• A virtual Ethernet adapter

Note that both physical and virtual I/O resources can be assigned to an IBM i virtual client partition. If a physical network adapter was not assigned to the IBM i client partition when it was first created, see the topic Managing physical I/O devices and slots dynamically using the HMC in the Logical Partitioning Guide (http://publib.boulder.ibm.com/infocenter/systems/scope/hw/topic/iphat/iphat.pdf available adapter.

Lastly, a virtual Ethernet adapter can also provide network connectivity to an IBM i client partition. To create one, consult the topic Configuring a virtual Ethernet adapter using the HMC in the Logical Partitioning Guide: http://publib.boulder.ibm.com/infocenter/systems/scope/hw/topic/iphat/iphat.pdf

) to assign an

If the IBM i client partition is using a virtual Ethernet adapter for networking, an SEA must be created in VIOS to bridge the internal virtual LAN (VLAN) to the external LAN. Use the HMC and the instructions in section 8.4 to perform the SEA configuration.

11.2. How to perform IBM i operator panel functions If the system is HMC-managed, follow these steps:

• Sign onto the HMC with a profile with sufficient authority to manage the IBM i client partition

• Select the partition

• Use the open-in-context arrow to select Serviceability Æ Control Panel Functions, then

the desired function.

If the system is IVM-managed, follow these steps:

• In IVM, click View/Modify Partitions

• Select the IBM i partition

• Use the More Tasks drop-down menu and select Operator panel service functions

• Select the function you wish to perform and click OK

11.3. How to display the IBM i partition System Reference Code (SRC) history

If the system is HMC-managed, follow these steps:

• Sign onto the HMC with a profile with sufficient authority to manage the IBM i client partition

• Select the partition

• Use the open-in-context arrow to select Serviceability Æ Reference Code History

• To display words 2 through 9 of a reference code, click the radio button for that code.

If the system is IVM-managed, follow these steps:

• In IVM, click View/Modify Partitions

• Select the IBM i partition

• Use the More Tasks drop-down menu and select Reference Codes

• Click an SRC to display all words.

11.4. Client IBM i LPARs considerations and limitations Consult the topic Considerations and limitations for i5/OS client partitions on systems

managed by the Integrated Virtualization Manager (IVM) in the Information Center:

http://publib.boulder.ibm.com/infocenter/systems/scope/i5os/topic/rzahc/rzahcbladei5limits.htm

11.5. Configuring Electronic Customer Support (ECS) over LAN

11.6. Backups

As mentioned above, IBM i as a client of VIOS can use a mix of virtual and physical I/O resources. Therefore, the simplest backup and restore approach is to assign an available tape adapter o n the system to it and treat it as a standard IBM i partition. The tape adapter can be any adapter supported by IBM i on IBM Power servers and can be shared with other partitions. To assign an available tape adapter to the IBM i client partition, consult the topic Managing physical I/O devices and slots dynamically using the HMC in the Logical Partitioning Gui de: (http://publib.boulder.ibm.com/infocenter/systems/scope/hw/topic/iphat/iphat.pdf

Once a tape adapter connected to a tape drive or library is available to the client partition, use the Backup and Recovery topic in the Information Center to manage backups: http://publib.boulder.ibm.com/infocenter/systems/scope/i5os/index.jsp?topic=/rzahg/rzahgbackup. htm&tocNode=int_215989.

12. DS4000 and DS5000 Copy Services and IBM i

IBM has conducted some basic functional testing of DS4000 and DS5000 and Copy Services with IBM i as client of VIOS. Below, you will find information on the scenarios tested and the resulting statements of support for using DS4000 and DS5000 Copy Services with IBM i.

12.1. FlashCopy and VolumeCopy

12.1.1. Test scenario

The following diagram shows the test environment used for FlashCopy and VolumeCopy:

12.1.2. FlashCopy and VolumeCopy support statements

The use of DS4000 and DS5000 FlashCopy and VolumeCopy with IBM i as a client of VIOS is supported as outlined below. Please note that to implement and use this solution, multiple

manual steps on the DS4000 or DS5000 storage subsystem, in VIOS and in IBM i are required. Currently, no toolkit exists that automates this solution and it is not part of IBM

PowerHA for IBM i. The components of the solution – DS4000 or DS5000 FlashCopy/VolumeCopy, VIOS and IBM i – must be managed separately and require the corresponding skill set. Note also that support for this solution will be provided by multiple IBM support organizations and not solely by the IBM i Support Center.

Support statements:

• DS4000 and DS5000 FlashCopy and VolumeCopy are supported by IBM i as a client of VIOS on both IBM Power servers and IBM Power blades

• Full-system FlashCopy and VolumeCopy when the production IBM i logical partition (LPAR) is powered off are supported

• Full-system FlashCopy and VolumeCopy when the production IBM i LPAR is in restricted state are supported

• The DS4000 and DS5000 'disable' and 're-create' functions with full-system FlashCopy and VolumeCopy when the production IBM i LPAR is powered off or is in restricted state are supported

• Full-system FlashCopy and VolumeCopy of the production IBM i logical partition (LPAR) after only using the IBM i 6.1 memory flush to disk (quiesce) function are not supported

• Full-system FlashCopy and VolumeCopy when the production IBM i LPAR is running are not supported

• FlashCopy and VolumeCopy of Independent Auxiliary Storage Pools (IASPs) are not supported

• Having the production and backup IBM i LPAR under the same VIOS is not supported

For assistance with using DS4000 and DS5000 FlashCopy and VolumeCopy with IBM i, contact IBM Systems Lab Services and Training using this Web site: http://www.ibm.com/systems/services/labservices/contact.html

12.2 Enhanced Remote Mirroring (ERM)

12.2.1. Test scenario

The following diagram shows the test environment used for ERM:

12.2.2 ERM support statements

The use of DS4000 and DS5000 Enhanced Remote Mirroring with IBM i as a client of VIOS is supported as outlined below. Please note that to implement and use this solution, multiple

manual steps on the DS4000 or DS5000 storage subsystem, in VIOS and in IBM i are required. Currently, no toolkit exists that automates this solution and it is not part of IBM

PowerHA for IBM i. The components of the solution – DS4000 or DS5000 ERM, VIOS and IBM i – must be managed separately and require the corresponding skill set. Note also that support for this solution will be provided by multiple IBM support organization and not solely by the IBM i Support Center.

IBM PowerHA for IBM i is also supported for IBM i as a client of VIOS. PowerHA for IBM i provides an automated, IBM i-driven replication solution that allows clients to leverage their existing IBM i skills. PowerHA for IBM i is supported by the IBM i Support Center. For more information on IBM PowerHA for IBM i, consult: http://www-

304.ibm.com/jct03001c/systems/power/software/availability/i5os.html. Support statements for ERM:

• DS4000 and DS5000 ERM is supported by IBM i as a client of VIOS on both IBM Power servers and IBM Power blades

• Synchronous ERM (DS4000 and DS5000 Metro Mirror) is supported

• Asynchronous ERM with Write Consistency Groups (DS4000 and DS5000 Global

Mirror) is supported

• Asynchronous ERM (DS4000 and DS5000 Global Copy) is not supported

• Full-system ERM (Metro and Global Mirror) for a planned switchover (IBM i production

LPAR is powered off) is supported

• Full-system ERM (Metro and Global Mirror) for an unplanned failover (IBM i production LPAR is running) is supported

• Replication of IASPs is not supported

• Suspending replication and accessing the replicated LUNs on backup site is not

supported

For assistance with using DS4000 and DS5000 Enhanced Remote Mirro ring with IBM i, contact IBM Lab Services using this Web site: http://www.ibm.com/systems/services/labservices/contact.html

13. IBM i using SAN Volume Controller (SVC) storage through VIOS

13.1. IBM i and SVC concepts

In October 2008, IBM announced SVC support for IBM i 6.1 as a client of VIOS. The ability to use storage virtualized by SVC further expands the IBM i open storage portfolio and allows clients to leverage their existing SVC infrastructure for IBM i. As mentioned previously, the support statement for IBM i as a client of VIOS includes all storage subsystems the SVC supports.

The concepts of using SVC storage for IBM i with VIOS are the same as those for other open storage. It is not possible to attach IBM i directly to SVC. Instead, 512-byte-per-sector virtual disks (VDisks) are created in SVC and physically connected to VIOS over the SAN fabric. SVC VDisks are recognized in VIOS as hdisks, or physical volumes, similar to LUNs from other open storage subsystems. Available hdisks are then assigned to an IBM i partition by mapping them to a Virtual SCSI server adapter (vhost) in VIOS, which is in turn connected to a Virtual SCSI client

adapter in IBM i. Once available to the IBM i client partition, SVC VDisks are used and managed no differently from integrated disk units or LUNs. They are assigned a disk type of 6B22, identifying them as virtual disk units provided by VIOS. For a detailed discussion of using virtual storage through VIOS, see section 8 of this document.

13.2. Attaching SVC storage to IBM i

The general process of attaching SVC storage to IBM i follows the one for using other open storage:

• SVC configuration

• VIOS installation and configuration

• IBM i installation and configuration

The SVC configurations steps are no different from virtualizing storage to a standard AIX host system. VIOS is installed and configured as for any other client partition, such as AIX or Linux. Existing skills in those areas continue to be applicable when making storage available to IBM i. There are no special versions of SVC and VIOS or additional software necessary for IBM i as a client of VIOS.

Before configuring new storage on SVC, review the planning and performance considerations in chapters 3 and 4 of the Redbook Implementing the IBM System Storage SAN Volume Controller V4.3 (SG246423), available at: http://www.redbooks.ibm.com/abstracts/sg246423.html?Open To create VDisks for IBM i on SVC and connect them over the SAN fabric to VIOS, consult chapters 8, and 9 or 10 of the same Redbook, using the instructions for AIX hosts.

For VIOS installation and configuration, consult the PowerVM Editions Guide, available at: http://publib.boulder.ibm.com/infocenter/systems/scope/hw/index.jsp?topic=/arecu/arecukickoff.ht m.

14. SVC Copy Services and IBM i IBM has conducted some basic functional testing of SVC Copy Services with IBM i as client of

VIOS. Below, you will find information on the scenarios tested and the resulting statements of support for using SVC Copy Services with IBM i.

14.1. FlashCopy

14.1.1. Test scenario

The following diagram shows the test environment used for FlashCopy:

14.1.2. FlashCopy statements The use of SVC FlashCopy with IBM i as a client of VIOS is supported as outlined below. Please

note that to implement and use this solution, multiple manual steps in SVC, in VIOS and in IBM i are required. Currently, no toolkit exists that automates this solution and it is not part of

IBM PowerHA for IBM i. The components of the solution – SVC FlashCopy, VIOS and IBM i – must be managed separately and require the corresponding skill set. Note also that suppo rt for this solution will be provided by multiple IBM support organizations and not solely by the IBM i Support Center.

Support statements:

• SVC FlashCopy is supported by IBM i as a client of VIOS on both IBM Power servers and IBM Power blades

• Full-system FlashCopy when the production IBM i logical partition (LPAR) is powered off is supported

• Full-system FlashCopy when the production IBM i LPAR is in restricted state is supported

• Full-system FlashCopy of the production IBM i logical partition (LPAR) after using the

IBM i 6.1 memory flush to disk (quiesce) function is supported

• Full-system FlashCopy when the production IBM i LPAR is running is not supported

• FlashCopy of Independent Auxiliary Storage Pools (IASPs) is not supported

• FlashCopy between VDisks using heterogeneous storage subsystems is supported

For assistance with using SVC FlashCopy with IBM i, contact IBM Lab Services using this Web site: http://www.ibm.com/systems/services/labservices/contact.html

14.2. Metro and Global Mirror

14.2.1. Test scenario

The following diagram shows the test environment used for Metro and Global Mirro r:

14.2.2. Metro and Global Mirror support statements

The use of SVC Metro and Global Mirror with IBM i as a client of VIOS is supported as outlined below. Please note that to implement and use this solution, multiple manual steps in SVC, in VIOS and in IBM i are required. Currently, no toolkit exists that automates this solution and it is not part of IBM PowerHA for IBM i. The components of the solution – SVC Metro or Global Mirror, VIOS and IBM i – must be managed separately and require the corresponding skill set. Note also that support for this solution will be provided by multiple IBM support organization and not solely by the IBM i Support Center.

304.ibm.com/jct03001c/systems/power/software/availability/i5os.html. Support statements for SVC Metro and Global Mirror:

• Both SVC Metro and Global Mirror are supported by IBM i as a client of VIOS on both IBM Power servers and IBM Power blades

• Only full-system replication is supported

• Replication of IASPs with Metro or Global Mirror is not supported

• Both Metro and Global Mirror are supported for a planned switchover (IBM i production

partition is powered off)

• Both Metro and Global Mirror are supported for an unplanned failover (IBM i production partition is running)

• Suspend and resume in case of link failure between SVC clusters is supported for both Metro and Global Mirror

For assistance with using SVC Metro and Global Mirror with IBM i, contact IBM Lab Services using this Web site: http://www.ibm.com/systems/services/labservices/contact.html

15. XIV Copy Services and IBM i IBM has conducted some basic functional testing of XIV Copy Services with IBM i as client of

VIOS. Below, you will find the resulting statements of support for using XIV Copy Services with IBM i.

Snapshots and Volume Copy:

• XIV snapshots and Volume Copy are supported by IBM i as a client of VIOS on both IBM Power servers and IBM Power blades

• Full-system snapshots and Volume Copy when the production IBM i logical partition (LPAR) is powered off are supported

• Full-system snapshots and Volume Copy when the production IBM i LPAR is in restricted state are supported

• Full-system snapshots and Volume Copy of the production IBM i logical partition (LPAR) after using the IBM i 6.1 memory flush to disk (quiesce) function are supported

• Full-system snapshots and Volume Copy when the production IBM i LPAR is running are not supported

• Snapshots and Volume Copy of Independent Auxiliary Storage Pools (IASPs) are not supported

Remote Mirror:

• XIV Remote Mirror (synchronous replication) is supported by IBM i as a client of VIOS on both IBM Power servers and IBM Power blades

• XIV asynchronous replication is also supported by IBM i as a client of VIOS on both IBM Power servers and IBM Power blades

• Only full-system replication is supported

• Replication of IASPs with Remote Mirror is not supported

• Remote Mirror is supported for a planned switchover (IBM i production partition is

powered off)

• Remote Mirror is supported for an unplanned failover (IBM i production partition is running)

16. DS5000 direct attachment to IBM i

16.1. Overview

As this document explains, IBM i has been able to connect to 512-byte-per-sector open storage through VIOS since early 2008 and the solution is implemented in many production environments. In October 2009, IBM further simplified open storage use for IBM i by announcing direct Fibre Channel (FC) attachment to the DS5100 and DS5300 subsystems. With the new enhancement, the IBM i LPAR owns one or more FC adapters, which are connected to the SAN fabric. 512B LUNs on the DS5100 or DS5300 are then mapped to the world-wide port names (WWPNs) of the FC adapter(s) in the IBM i LPAR. VIOS is no longer required in this case. Note that one or more VIOS LPARs may still exist on the same Power server to take advantages of other technologies, such as Active Memory Sharing (AMS), N_Port ID Virtualization (NPIV), Live Partition Mobility for AIX and Linux or other I/O virtualization. However, VIOS is no longer required in order for IBM i to use storage from DS5100 and DS5300. It is possible to have the same or another IBM i LPAR on the same Power server still use VIOS for optical virtualization, for example.

A note on terminology: while the new enhancement is referred to as “direct attachment of IBM i to DS5100 and DS5300,” IBM expects that in most cases the attachment will be through a FC switch. While directly attaching FC cables from the adapters on the Power server owned by IBM i to the host channel ports on the storage is supported, this approach decreases the flexibility of the SAN environment. Therefore, “direct attachment” in this case refers to “without VIOS.”

There are two main benefits of directly attaching IBM i to DS5100 and DS5300:

• VIOS skills are not required to implement and use the solution, only IBM i and SAN skills

• The processor, memory, I/O and software expenditure to run a VIOS LPAR is not

required

Multi-path I/O for IBM i in this environment is supported with a maximum of 8 paths. No additional software on the DS5100 or DS5300 or IBM i is required. The IBM i FC adapter driver employs the round-robin algorithm by default; it cannot be changed.

16.2. Supported hardware and software

As of the publication date, only the following hardware is supported for IBM i directly attaching to DS5100 and DS5300. This support is also reflected in section 7.1.

• DS5100 (machine-type 1818, model 51A)

• DS5300 (machine-type 1818, model 53A)

• EXP810 expansion unit

• EXP5000 expansion unit

• EXP5060 expansion unit

• Fibre Channel and SATA HDDs (hard disk drives), with FC drives recommended for

production workloads

• POWER6 processor-based or later Power servers

• FC #5749 4Gb dual-port PCI-X DDR Fibre Channel adapter

• FC #5774 4Gb dual-port PCI-E Fibre Channel adapter

• FC #5735 8Gb dual-port PCI-E Fibre Channel adapter

As of the publication date, only the following software or firmware is supported for IBM i directly attaching to DS5100 and DS5300. This support is also reflected in section 7.2.

• IBM i 6.1 with LIC 6.1.1 or later

• Controller firmware 07.60.28.00 or later

• DS Storage Manager 10.60.x5.17 or later

• IBM i Host Attachment Kit, FC #7735 (required)

• Storage Partitioning (strongly recommended)

16.3. Best practices, limitations and performance

As with other storage subsystems, it recommended to create LUNs for IBM i production workloads on FC drive RAID arrays only in order to meet the performance requirements of the application. While SATA drives are supported for direct attachment to IBM i, they are best suited for test, development and archival IBM i workloads. Mixing of FC and SATA drives within an enclosure is supported. Solid-state drives are not presently supported for IBM i direct attachment, but IBM i is investigating offering that support. Hardware encryption of the data on the physical drives within the storage subsystem is supported.

All LUNs mapped directly to IBM i must be protected. Therefore, the supported RAID levels for IBM i direct attachment are 1, 5 and 6. LUNs report as protected in IBM i; therefore, IBM i mirroring is not possible. The reliability of the solution is based on the RAID protection within the storage subsystem and on using multiple paths. The maximum LUN size for IBM i in this environment is up to but not including 2TB. A maximum of 64 LUNs per FC adapter port is supported; this statement is identical to that for DS8000 direct attachment. Dynamic volume (LUN) expansion is not supported for IBM i. Once LUNs have been mapped to IBM i, at most 300 from the same subsystem can be added to an Auxiliary Storage Pool (ASP) with a single operation. More than 300 LUNs can be added to an ASP in general, but they require multiple add operations.

There are no special requirements for the load source LUN, except of course sufficient size to qualify as a load source for IBM i 6.1. The load source does not need to be the very first LUN mapped to IBM i. When performing an Initial Program Load (IPL), an active path to the load source LUN is required. Path failover is not possible during IPL; however, once IPL is complete, failover functionality resumes.

As with other host types, each port on a Fibre Channel adapter in IBM i should be zoned separately to controller A or B on the storage subsystem. Furthermore, it is strongly recommended to use Storage Partitioning when directly connecting to IBM i. Storage Partitioning is a premium feature on the DS5100 and DS5300 which allows a set of LUNs to be associated only with specified hosts and host ports. Storage Partitioning allows the creation of multiple host groups; only the hosts within a particular host group are allowed access to a group of LUNs. Without that premium feature, all hosts reside in the Default Host Group, which allows IBM i to access LUNs mapped to other hosts. The recommended approach is to use Storage Partitioning and create a host group for each IBM i LPAR using the storage subsystem, placing the hosts associated with that LPAR in the host group. Hosts representing different IBM i LPARs should not be placed in the same host group, as IBM i does not support shared simultaneous access to the same LUNs from multiple LPARs.

Note that migrating from IBM i using DS5100 or DS5300 storage through VIOS to IBM i using the same subsystem directly requires data migration, typically by save and restore. There is no option to reassign the LUNs from a VIOS host to an IBM i host or use FlashCopy or VolumeCopy.

Performance for IBM i directly attached to DS5100 and DS5300 is very similar to that of the same adapter and storage configuration attached through VIOS. Direct attachment performance will not be significantly better than that of VIOS attachment for IBM i. Performance results for IBM i directly attached to DS5300 are in Chapter 14.5.3 of the Performance Capabilities Reference manual at: http://www.ibm.com/systems/i/advantages/perfmgmt/resource.html

16.4. Sizing and configuration

There are two main sources of sizing information when planning a configuration involving IBM i directly accessing DS5100 or DS5300: the Disk Magic sizing tool and the Performance Capabilities Reference manual mentioned in the previous chapter. It is recommended that they be used jointly. While there is not yet a direct attachment model for IBM i with DS5100 or DS5300 in Disk Magic, the existing VIOS model can still be used to determine the physical storage configuration, such as number of physical drives, number and type of RAID arrays and amount of cache. At the same time, the Performance Capabilities Reference manual offers some excellent suggestions on the number of LUNs and paths to use based on the I/Os per second and throughput required. Note that the optimal number of LUNs for IBM i direct attachment is significantly higher than that for VIOS attachment; typically 64 or 128 in the direct case.

Configuration of IBM i direct attachment to DS5100 and DS5300 can be divided into two general steps:

• Storage and SAN configuration

• Power server hardware and LPAR configuration

To configure the DS5100 or DS5300, use the Redbook Introduction to the IBM System Storage DS5000 Series (SG24-7676), available at:

http://www.redbooks.ibm.com/abstracts/sg247676.html?Open and limitations outlined in the previous section of this document. When creating a host for direct IBM i attachment on the storage subsystem, the host type should be “IBM i.” To configure the SAN fabric, follow the instructions from your FC switch manufacturer.

To create the IBM i LPAR and assign the FC adapter(s) to it, follow the instructions in the Logical Partitioning Guide, available at: http://publib.boulder.ibm.com/infocenter/systems/scope/hw/topic/iphat/iphat.pdf

16.5. Copy Services

IBM has conducted some basic functional testing of DS5100 and DS5300 Copy Services when directly attached to IBM i. Below, you will find the resulting support statements. Note that using Copy Services when directly attaching DS5100 and DS5300 involves manual steps in both the DS Storage Manager GUI and IBM i. PowerHA for IBM i does not support DS5100 and DS5300 Copy Services when directly attaching to IBM i (or through VIOS).

FlashCopy and VolumeCopy:

• FlashCopy and VolumeCopy are supported when directly attaching DS5100 and DS5300 to IBM i on POWER6 or later processor-based Power servers

• As mentioned earlier, Power blades are not supported for direct IBM i attachment to DS5100 and DS5300 in general

• Full-system FlashCopy and VolumeCopy when the production IBM i logical partition (LPAR) is powered off are supported

• Full-system FlashCopy and VolumeCopy when the production IBM i LPAR is in restricted state are supported

• Full-system FlashCopy and VolumeCopy of the production IBM i logical partition (LPAR) after using the IBM i 6.1 memory flush to disk (quiesce) function are not supported

• Full-system FlashCopy and VolumeCopy when the production IBM i LPAR is running are not supported

, while following the best practices

• FlashCopy and VolumeCopy of Independent Auxiliary Storage Pools (IASPs) are supported

Enhanced Remote Mirroring (ERM):

• DS5100 and DS5300 ERM with direct attachment to IBM i is supported on IBM Power servers (but not on IBM Power blades)

• Synchronous ERM (DS5100 and DS5300 Metro Mirror) is supported

• Asynchronous ERM with Write Consistency Groups (DS5100 and DS5300 Global

Mirror) is supported

• Asynchronous ERM (DS5100 and DS5300 Global Copy) is not supported

• Full-system ERM (Metro and Global Mirror) for a planned switchover (IBM i production

LPAR is powered off) is supported

• Full-system ERM (Metro and Global Mirror) for an unplanned failover (IBM i production LPAR is running) is supported

• ERM (Metro and Global Mirror) with IASPs for a planned switchover (IBM i production LPAR is powered off) is supported

• ERM (Metro and Global Mirror) with IASPs for an unplanned failover (IBM i production LPAR is running) is supported

17. N_Port ID Virtualization (NPIV) for IBM i

17.1. Overview

NPIV is an industry-standard Fibre Channel (FC) protocol that allows VIOS to directly share a single FC adapter among multiple client LPARs. Unlike Virtual SCSI (VSCSI), NPIV does not map a LUN to a virtual target device in VIOS, which the client LPAR can then access as a generic SCSI disk. Instead, a port on the physical FC adapter is mapped to a Virtual Fibre Channel (VFC) server adapter in VIOS, which in turn is connected to a VFC client adapter in IBM i. When the VFC client adapter is created, two unique world-wide port names (WWPNs) are generated for it. Through the link to the server VFC adapter and then the physical adapter in VIOS, those WWPNs become available on the SAN and storage can be mapped to them as with any other FC host ports. Note that these WWPNs are unique not just within the Power server, but globally on the SAN. The PowerVM Hypervisor creates two WWPNs per VFC client adapter to facilitate Live Partition Mobility for LPARs using NPIV. Live Partition Mobility is currently supported for AIX and Linux LPARs. When a VFC client adapter is deleted, the WWPNs are not reused. By default, the PowerVM Hypervisor is capable of creating 32,000 WWPNs for VFC client adapters. If additional WWPNs are required, clients can acquire an enablement code from IBM. The following diagram illustrates using NPIV for IBM i:

From the storage subsystem’s perspective, the LUNs, volume group and host connection are created as though the IBM i LPAR is directly connected to the storage through the SAN fabric. While VIOS still plays a role with NPIV, that role is much more of a passthrough one compared to VSCSI. Note that an 8Gb FC adapter is required; however, the FC switches do not need to be 8Gb. Additionally, only the first switch – the one to which the FC adapter on the Power server is directly connected – must be NPIV-capable. The rest of the SAN fabric does not need to be NPIV-capable.

17.2. Supported hardware and software As of the publication date, NPIV for IBM i supports only the following hardware and software:

• POWER6 processor-based or later Power servers

• The FC #5735 8Gb dual-port PCI-E Fibre Channel adapter

• The DS8000 storage subsystem series

• The following tape libraries with LTO4 drives only:

o TS3100 (machine-type 3573) o TS3200 (M/T 3573) o TS3500 (M/T 3584)

• IBM i 6.1 with LIC 6.1.1 or later

• VIOS 2.1.2 or later

17.3. Configuration

There are two general steps in configuring NPIV for IBM i:

• LPAR and VIOS configuration on the Power server

• Storage subsystem or tape library configuration

To perform the LPAR and VIOS setup, consult Chapter 2.9 in the Redbook PowerVM Managing and Monitoring (SG 24-7590): http://www.redbooks.ibm.com/abstracts/sg247590.html?Open.

While the examples given are for an AIX client of VIOS, the procedure is identical for an IBM i client.

To perform the storage or tape configuration, consult the Redbook IBM i and IBM System Storage: A Guide to Implementing External Disks on IBM i (SG24-7120) at http://www.redbooks.ibm.com/abstracts/sg247120.html?Open (SG 24-7440) at http://www.redbooks.ibm.com/abstracts/sg247440.html?Open above, from the storage subsystem’s or tape library’s perspective, the configuration is identical to that for IBM i directly attached through the SAN fabric. Accordingly, DS8000 LUNs may be created as “IBM i protected” or “IBM i unprotected” and will correctly report as such to Storage Management in IBM i. A tape library and a drive within will also report the correct device names and types, such as TAPMLBxx, 3584 and TAPxx, 3580-004. All tape operations supported with direct FC attachment are supported through NPIV, including hardware tape encryption.

17.4. Copy Services

As explained above, the main differences between NPIV and VSCSI I/O virtualization for IBM i are:

• With NPIV, storage and tape devices report into IBM i with their specific characteristics, as opposed to generic SCSI devices

• IBM i has also gained NPIV access to storage and tape devices to which it could already attached directly, benefiting from transparent FC adapter sharing

As a result, both the storage subsystem (DS8000) and LUN types involved with NPIV for IBM i are identical to those already supported by PowerHA for IBM i. PowerHA for IBM i continues to manage both Copy Services on DS8000 using the Storage Management Console (SMC) and the required clustering objects in IBM i from a single interface. Therefore, DS8000 Copy Serv ices are supported by PowerHA for IBM i when using NPIV, including FlashCopy, Metro and Global Mirror.

or Implementing IBM Tape in i5/OS

. As mentioned

18. Additional resources

18.1. IBM i

• IBM i on a Power Blade Read-me First: http://www.ibm.com/systems/power/hardware/blades/ibmi.html

• IBM STG Lab Services: http://www.ibm.com/systems/services/labservices/contact.html

• Logical Partitioning Guide:

http://publib.boulder.ibm.com/infocenter/systems/scope/hw/topic/iphat/iphat.pdf

• IBM i installation: http://publib.boulder.ibm.com/infocenter/systems/scope/i5os/index.jsp?topic=/rzahc/rzahc

1.htm&tocNode=int_216451

• IBM i PTF installation: http://publib.boulder.ibm.com/infocenter/systems/scope/i5os/index.jsp?topic=/rzam8/rzam 8fixconcepts1.htm

• Latest recommended IBM i PTFs: http://www-

912.ibm.com/s_dir/slkbase.nsf/recommendedfixes

• IBM Fix Central: http://www-912.ibm.com/eserver/support/fixes/

• IBM Systems Workload Estimator: http://www-

304.ibm.com/systems/support/tools/estimator/index.html

• IBM i networking: http://publib.boulder.ibm.com/infocenter/systems/scope/i5os/topic/rzajy/rzajyoverview.htm

• ECS over LAN: http://publib.boulder.ibm.com/infocenter/systems/scope/i5os/topic/rzaji/rzaji_setup.htm

• IBM i virtual client partitions topics in the IBM i Information Center: http://publib.boulder.ibm.com/infocenter/systems/scope/i5os/topic/rzahc/rzahcbladei5limit s.htm

• IBM PowerHA for IBM i: http://www-

304.ibm.com/jct03001c/systems/power/software/availability/i5os.html.

18.2. Storage

• IBM System Storage DS4000 and Storage Manager V10.10 (Redbook): http://www.redbooks.ibm.com/redpieces/abstracts/sg247010.html?Open

• DS4000 Web site: http://www.ibm.com/systems/storage/disk/ds4000/index.html

• Performance Capabilities Reference manual (chapter 14.5 for best practices for IBM i

and VIOS): http://www.ibm.com/systems/i/solutions/perfmgmt/resource.html

• SDD-PCM driver: http://www.ibm.com/support/docview.wss?rs=540&context=ST52G7&dc=D430&uid=ssg1 S4000201&loc=en_US&cs=utf-8&lang=en

• IBM Redbooks site: http://www.redbooks.ibm.com

• IBM i and IBM System Storage: A Guide to Implementing External Disks on IBM i

(Redbook): http://www.redbooks.ibm.com/abstracts/sg247120.html?Open

• Implementing IBM Tape in i5/OS (Redbook): http://www.redbooks.ibm.com/abstracts/sg247440.html?Open

• Introduction to the IBM System Storage DS5000 Series (Redbook): http://www.redbooks.ibm.com/abstracts/sg247676.html?Open

18.3. VIOS

• VIOS Web site: http://www14.software.ibm.com/webapp/set2/sas/f/vios/home.html

• PowerVM Editions Guide:

http://publib.boulder.ibm.com/infocenter/systems/scope/hw/index.jsp?topic=/arecu/arecuk ickoff.htm

. . .

• Advanced POWER Virtualization on IBM System p5: Introduction and Configuration (Redbook): http://www.redbooks.ibm.com/abstracts/sg247940.html?Open

• VIOS command reference: http://publib.boulder.ibm.com/infocenter/systems/scope/hw/index.jsp?topic=/iphb1/iphb1_ vios_commandslist.htm

• VIOS Datasheet: http://www14.software.ibm.com/webapp/set2/sas/f/vios/documentation/datasheet.html

• DLPAR Checklist: http://www.ibm.com/developerworks/systems/articles/DLPARchecklist.html

• PowerVM Managing and Monitoring (Redbook): http://www.redbooks.ibm.com/abstracts/sg247590.html?Open

18.4. SVC

• SVC overview Web site: http://www.ibm.com/systems/storage/software/virtualization/svc

• Interoperability information:

http://www.ibm.com/systems/storage/software/virtualization/svc/interop.html

• Implementing the IBM System Storage SAN Volume Controller V4.3 (Redbook): http://www.redbooks.ibm.com/abstracts/sg246423.html?Open

• SAN Volume Controller Best Practices and Performance Guidelines (Redbook): http://www.redbooks.ibm.com/abstracts/sg247521.html?Open

• SVC V4.3.0 Advanced Copy Services (Redbook): http://www.redbooks.ibm.com/redpieces/abstracts/sg247574.html?Open

• IBM SAN Volume Controller 4.2.1 Cache Partitioning (Redpiece): http://www.redbooks.ibm.com/abstracts/redp4426.html?Open

. .

19. Trademarks and disclaimers

This document was developed for IBM offerings in the United States as of the date of publication. IBM may not make these offerings available in other countries, and the information is subject to change without notice. Consult your local IBM business contact for information on the IBM offerings available in your area. Information in this document concerning non-IBM products was obtained from the suppliers of these products or other public sources. Questions on the capabilities of non-IBM products should be addressed to the suppliers of those products. IBM may have patents or pending patent applications covering subject matter in this document. The furnishing of this document does not give you any license to these patents. Send license inquires, in writing, to IBM Director of Licensing, IBM Corporation, New Castle Drive, Armonk, NY 10504-1785 USA. All statements regarding IBM future direction and intent are subject to change or withdra wal without notice, and represent goals and objectives only. The information contained in this document has not been submitted to any formal IBM test and is provided "AS IS" with no warranties or guarantees either expressed or implied. All examples cited or described in this document are presented as illustrations of the manner in which some IBM products can be used and the results that may be achieved. Actual environmental costs and performance characteristics will vary depending on individual client configurations and conditions. IBM Global Financing offerings are provided through IBM Credit Corporation in the United States and other IBM subsidiaries and divisions worldwide to qualified commercial and government clients. Rates are based on a client's credit rating, financing terms, offering type, equipment type and options, and may vary by country. Other restrictions may apply. Rates and offerings are subject to change, extension or withdrawal without notice. IBM is not responsible for printing errors in this document that result in pricing or information inaccuracies. All prices shown are IBM's United States suggested list prices and are subject to change without notice; reseller prices may vary. IBM hardware products are manufactured from new parts, or new and serviceable used parts. Regardless, our warranty terms apply. Any performance data contained in this document was determined in a controlled environment. Actual results may vary significantly and are dependent on many factors including system hardware configuration and software design and configuration. Some mea sure ments quoted in this document may have been made on development-level systems. There is no guarantee these measurements will be the same on generally-available systems. Some measurements quoted in this document may have been estimated through extrapolation. Users of this document should verify the applicable data for their specific environment.

The following terms are registered trademarks of International Business Machines Corporation in the United States and/or other countries: AIX, AIX/L, AIX/L (logo), AIX 6 (logo), alphaWorks, AS/400, BladeCenter, Blue Gene, Blue Lightning, C Set++, CICS, CICS/6000, ClusterProven, CT/2, DataHub, DataJoiner, DB2, DEEP BLUE, developerWorks, DirectTalk, Domino, DYNIX, DYNIX/ptx, e business (logo), e(logo)business, e(logo)server, Enterprise Storage Server, ESCON, FlashCopy, GDDM, IBM i, IBM i (logo), IBM, IBM (logo), ibm.com, IBM Business Partner (logo), Informix, IntelliStation, IQ-Link, LANStreamer, LoadLeveler, Lotus, Lotus Notes, Lotusphere, Magstar, MediaStreamer, Micro Channel, MQSeries, Net.Data, Netfinity, NetView, Network Station, Notes, NUMA-Q, OpenPower, Operating System/2, Operating System/400, OS/2, OS/390, OS/400, Parallel Sysplex, PartnerLink, PartnerWorld, Passport Advantage, POWERparallel, Power PC 603, Power PC 604, PowerPC, PowerPC (logo), Predictive Failure Analysis, pSeries, PTX, ptx/ADMIN, Quick Place, Rational, RETAIN, RISC System/6000, RS/6000, RT Personal Computer, S/390, Sametime, Scalable POWERparallel Systems, SecureWay, Sequent, ServerProven, SpaceBall, System/390, The Engines of ebusiness, THINK, Tivoli, Tivoli (logo), Tivoli Management Environment, Tivoli Ready (logo), TME, TotalStorage, TURBOWAYS, VisualAge, WebSphere, xSeries, z/OS, zSeries.

The following terms are trademarks of International Business Machines Corporation in the United States and/or other countries: Advanced Micro-Partitioning, AIX 5L, AIX PVMe, AS/400e, Calibrat ed Vectored Cooling, Chiphopper, Chipkill, Cloudscape, DataPower, DB2 OLAP Server, DB2 Universal Data base,

DFDSM, DFSORT, DS4000, DS6000, DS8000, e-business (logo), e-business on demand, EnergyScale, Enterprise Workload Manager, eServer, Express Middleware, Express Portfolio, Express Servers, Express Servers and Storage, General Purpose File System, GigaProcessor, GPFS, HACMP, HACMP/6000, IBM Systems Director Active Energy Manager, IBM TotalStorage Proven, IBMLink, IMS, Intelligent Miner, iSeries, Micro-Partitioning, NUMACenter, On Demand Business logo, POWER, PowerExecutive, PowerVM, PowerVM (logo), Power Architecture, Power Everywhere, Power Family, POWER Hypervisor, Power PC, Power Systems, Power Systems (logo), Power Systems Software, Power Systems Software (logo), PowerPC Architecture, PowerPC 603, PowerPC 603e, PowerPC 604, PowerPC 750, POWER2, POWER2 Architecture, POWER3, POWER4, POWER4+, POWER5, POWER5+, POWER6, POWER6+, pure XML, Quickr, Redbooks, Sequent (logo), SequentLINK, Server Advantage, ServeRAID, Service Director, SmoothStart, SP, System i, System i5, System p, System p5, System Storage, System z, System z9, S/390 Parallel Enterprise Server, Tivoli Enterprise, TME 10, TotalStorage Proven, Ultramedia, VideoCh arger, Virtualization Engine, Visualization Data Explorer, Workload Partitions Manager, X-Architecture, z/Architecture, z/9.

A full list of U.S. trademarks owned by IBM may be found at: http://www.ibm.com/legal/copytrade.shtml The Power Architecture and Power.org wordmarks and the Power and Power.org logos and related marks are trademarks and service marks licensed by Power.org. UNIX is a registered trademark of The Open Group in the United States, other countries or both. Linux is a trademark of Linus Torvalds in the United States, other countries or both. Microsoft, Windows, Windows NT and the Windows logo are registered trademarks of Microsoft Corporation in the United States, other countries or both. Intel, Itanium, Pentium are registered trademarks and Xeon is a trademark of Intel Corporation or its subsidiaries in the United States, other countries or both. AMD Opteron is a trademark of Advanced Micro Devices, Inc. Java and all Java-based trademarks and logos are trademarks of Sun Microsystems, Inc. in the United States, other countries or both. TPC-C and TPC-H are trademarks of the Transaction Performance Processing Council (TPPC). SPECint, SPECfp, SPECjbb, SPECweb, SPECjAppServer, SPEC OMP, SPECviewperf, SPECapc, SPEChpc, SPECjvm, SPECmail, SPECimap and SPECsfs are trademarks of the Standard Performance Evaluation Corp (SPEC). NetBench is a registered trademark of Ziff Davis Media in the United States, other countries or both. AltiVec is a trademark of Freescale Semiconductor, Inc. Cell Broadband Engine is a trademark of Sony Computer Entertainment Inc. InfiniBand, InfiniBand Trade Association and the InfiniBand design marks are trademarks and/or service marks of the InfiniBand Trade Association. Other company, product and service names may be trademarks or service marks of others.

Ibm OPEN STORAGE READ ME FIRST 7-9-2010, I VIRTUALIZATION READ ME FIRST 7-9-2010 User Manual

Specifications and Main Features

Frequently Asked Questions

User Manual