Implementing a Virtual Server Environment:
Getting Started
Chapter 1: Introduction ....................................................................................................................... 3
Chapter 2: Assessing and Planning....................................................................................................... 4
Long-term roadmap .......................................................................................................................... 4
Short-term roadmap.......................................................................................................................... 4
Chapter 3: Understanding the Choices for Virtualization Technologies...................................................... 6
Partitioning Solutions: ....................................................................................................................... 6
Why choose nPartitions (nPars)? ........................................................................................................ 6
Key Benefits................................................................................................................................. 6
Trade-offs .................................................................................................................................... 6
Sweet Spots................................................................................................................................. 6
Why choose Virtual Partitions (vPars)? ................................................................................................ 6
Key Benefits................................................................................................................................. 6
Trade-offs .................................................................................................................................... 7
Sweet Spots................................................................................................................................. 7
Why choose Integrity Virtual Machines (VMs)? .................................................................................... 7
Key Benefits................................................................................................................................. 7
Trade-offs .................................................................................................................................... 7
Sweet Spots................................................................................................................................. 8
Why choose Resource Partitions or Secure Resource Partitions (SRPs)?.................................................... 8
Key Benefits................................................................................................................................. 8
Trade-offs .................................................................................................................................... 8
Sweet Spots................................................................................................................................. 9
HP Utility Pricing Solutions:................................................................................................................ 9
Why choose Instant Capacity (iCAP)? ................................................................................................ 9
Key Benefits................................................................................................................................. 9
Trade-offs .................................................................................................................................... 9
Sweet Spots................................................................................................................................. 9
Why choose Temporary Instant Capacity (TiCAP)?............................................................................... 9
Key Benefits................................................................................................................................. 9
Trade-offs .................................................................................................................................... 9
Sweet Spots................................................................................................................................. 9
Why choose Global Instant Capacity (GiCAP)................................................................................... 10
Key Benefits............................................................................................................................... 10
Trade-offs .................................................................................................................................. 10
Sweet Spots............................................................................................................................... 10
Automation Solutions:..................................................................................................................... 10
Why choose Workload Manager (WLM) or Global Workload Manager (gWLM)?................................ 10
Key Benefits............................................................................................................................... 10
Trade-offs .................................................................................................................................. 10
Sweet Spots............................................................................................................................... 10
Chapter 4: HP VSE Reference Architectures ......................................................................................... 12
Chapter 5: Identifying a Pilot Project................................................................................................... 13
General guidelines......................................................................................................................... 13
Chapter 6: Making Your Choices ....................................................................................................... 14
Determine CPU processing requirements for workloads....................................................................... 14
Capacity Advisor ........................................................................................................................... 15
Choose the type of partitioning........................................................................................................ 15
Choose the type of Integrity server ................................................................................................... 16
Utility Pricing ................................................................................................................................. 17
Organizing partitions and resource sharing ...................................................................................... 17
Automation ................................................................................................................................... 18
High Availability and Disaster Tolerance........................................................................................... 18
Implement the Pilot Project............................................................................................................... 19
Repeat the Process ......................................................................................................................... 19
For more information.......................................................................................................................... 20
Chapter 1: Introduction
Many customers have said that they like the concept of the HP Virtual Server Environment (VSE), but
that it seems complex and disruptive to their environment, and they don’t know where to begin.
Achieving a virtualized environment and a truly Adaptive Infrastructure is a journey, and not an all-ornothing goal that must be reached in the first implementation. You do not even need a completely
defined VSE architecture in place in order to start taking advantage of the many benefits offered by
HP VSE. You can start benefiting from virtualization technologies without introducing disruptive and
time-consuming implementations. One great thing about HP’s virtualization technologies is that they
have all been designed to work well together. You can pick and choose which technologies are right
for your business needs – both now and in the future -- and be confident that they will work together.
Note:
This paper is intended for readers who are familiar with the HP
virtualization technologies that comprise the HP Virtual Server Environment.
References to more information, including a white paper entitled, “An
Introduction to the HP Virtual Server Environment “can be found on the last
page of this paper.
3
Chapter 2: Assessing and Planning
Every organization is unique and has different challenges and problems to solve, but all must use a
similar process. This chapter is intended to help you organize your thoughts and plans.
Long-term roadmap
The first step is to identify your organization’s business drivers and long-term goals. For most
customers, the ultimate goal is to create an agile and responsive IT organization, one that aligns with
the needs of the business and, therefore, is both a service to the business and a competitive
advantage. For this reason, it is important to understand the key business drivers and to think about
what you want your environment to look like in the long term. The following are some pertinent
questions and considerations:
• Analyze your application environment and some of your business processes as they relate to
the IT organization. Do you have more than one line of business, and does each line of
business or application group own their own servers for testing and production? Application
“silos” such as this are a leading cause of underutilized servers and server sprawl. One longterm goal might be to consolidate these applications and allow IT to provision the necessary
resources as they are needed. This approach can allow the application groups to focus on
meeting the needs of the businesses they support instead of worrying about the IT
infrastructure.
• Are you expecting significant growth in the business or growth through acquisitions or
mergers? If so, is it important to be able to deploy new applications or new instances of
existing applications quickly? Some HP VSE customers find that a virtualized environment
allows them to deploy a new application in days rather than months.
• Do you need to achieve a 24x7 environment or add a disaster-recovery capability? HP
virtualization technologies are well integrated with the HP Serviceguard suite of highavailability and disaster-tolerant solutions.
• Do you have to support and maintain many different versions of the operating system and
application software? Does your organization already have certain standardized processes,
or is each application group completely independent? If you consolidate and share
resources, what kinds of internal or “political” issues will you need to address? These issues
might not be trivial to solve, but it is important to understand them. You can address many
problems and pain points using the virtualization technologies within these “application
silos,” but in the long run, customers who can also solve the business process issues will
benefit the most from a Virtual Serve Environment and can achieve a more robust Adaptive
Infrastructure.
Short-term roadmap
As previously described, defining a long-term architecture for a virtualized environment requires
significant thought and planning, and perhaps even some significant changes in your IT infrastructure
and business processes. The good news is that the VSE architecture does not have to be completely
defined before you can take advantage of HP virtualization technologies. You may need to address
some pain points and solve some short term problems first.
It is clear that most companies need to reduce their infrastructure costs. The problems of
overprovisioning, underutilized servers, server sprawl, paying too much for software licenses and
support, increasing cost of power and cooling, and simply running out of space in the data center are
4
common. Which of these problems apply to your organization, and which ones need to be
addressed first?
For most customers, any change in the IT infrastructure or application environment poses some level of
complexity or risk. This paper discusses the various ways that you can implement changes
incrementally, from those that are least disruptive to your current environment to those that represent
the greatest change and therefore provide the greatest benefit.
5
Chapter 3: Understanding the Choices for Virtualization Technologies
This chapter describes the key benefits, trade-offs, and sweet spots for some of the HP virtualization
technologies. Understanding these will help you determine which technology is most appropriate for
solving a specific problem or achieving a certain benefit. Use this information as you decide on a
pilot project and assess the level of complexity and the amount of change that may be required to
implement them.
Note:
Most of the information in this chapter is taken from a book entitled The HP
Virtual Server Environment. For information about how to obtain this book,
see the last page of this paper.
Partitioning Solutions:
Why choose nPartitions (nPars)?
Key Benefits
• Hardware fault isolation (electrical).
• Operating system isolation.
• Choice of OS (HP-UX, Linux, Windows®, OpenVMS).
• No negative performance impact (in some cases you might see improved performance
resulting from less SMP overhead).
• Easy implementation.
• Dynamic cell OLAR (Online Addition and Removal) with HP-UX 11i v3.
Trade-offs
• Requires cell-based system.
• Granularity for a partition is at the cell level.
• No resource sharing across nPars (unless being flexed with Instant Capacity cores).
Sweet Spots
• Mission-critical applications that require fault isolation and dedicated resources.
• Need to run multiple operating systems on the same physical server.
• nPars are supported on both PA-RISC and HP Integrity servers.
• nPars can be a mix of PA-RISC and HP Integrity on Superdome servers (excellent for mixed
environments or during transitions).
Why choose Virtual Partitions (vPars)?
Key Benefits
• Dedicated hardware resources (processor, memory, I/O).
• Partition size can be scaled in increments of 1 processor core.
6
• Operating system isolation (each vPar is a unique instance of HP-UX).
• CPU and memory resources can be changed or moved dynamically between vPars that are
within the same nPar.
• Negligible overhead
• Good choice for I/O-intensive applications as compared with HP Integrity Virtual Machines.
Trade-offs
• No hardware fault isolation for vPars running in the same nPar (that is, a hardware failure
within an nPar will affect all vPars in that nPar).
• HP-UX is the only operating system supported.
• Only supported on cell-based systems if using Integrity servers (some older, non-cell-based
systems are supported on PA-RISC).
• Requires dedicated hardware resources, which might result in overprovisioning for small
workloads.
Sweet Spots
• A good choice if you require finer granularity than an nPar but still need dedicated
hardware, have I/O-intensive applications, or need a unique instance or version of HP-UX.
• A good choice if you want to dynamically move processor or memory resources between
vPars (within the same nPar).
• Instant Capacity or Temporary Instant Capacity resources can be activated for any vPar
within the same nPar.
• Easy for deploying a new application or a new instance of an existing application by simply
creating a new vPar (instead of deploying a new server).
• Different versions of HP-UX (e.g. 11iv1, 11iv2, and 11iv3) can run in separate vPars within
the same nPar.
Why choose Integrity Virtual Machines (VMs)?
Key Benefits
• Granularity is sub-CPU (as little as 5%).
• Virtual CPUs (vCPUs), CPU entitlements, and memory can be changed dynamically.
• Dedicated hardware not required; CPU and I/O resources are shared.
• Supported on all Integrity systems (running HP-UX 11i v2 or later) and on both cell-based and
non-cell-based servers, including Integrity server blades.
• Operating system isolation and flexibility (that is, each guest OS is a unique instance).
• Multiple OS guests are supported (HP-UX, Windows, and Linux; OpenVMS is planned).
• OS guests can run without modification.
• Any virtual storage device (disk, CD, DVD) can be implemented as a file.
• An ISO image of a preferred software image can be implemented as a virtual DVD, and can
be used to quickly deploy common software, operating system updates, or patch bundles to
multiple virtual machines.
• Starting with the 4.1 version, they can be migrated online from one host to another.
Trade-offs
• No hardware fault isolation.
7
• Integrity VMs and vPars cannot be used within the same nPar.
• Each Integrity VM currently limited to 8 cores.
• Hardware resources are shared, so not a good choice if dedicated hardware is required.
• Not supported on PA-RISC systems.
• There is a slight decrease in performance for I/O, so not the best choice for I/O-intensive
applications. Note that newer releases of Integrity Virtual Machines deliver improved I/O
performance using Accelerated Virtual I/O drivers that streamline and re-architect the I/O
path for both networking and disk I/O.
Sweet Spots
• Good choice for applications that do not need dedicated hardware (or an entire CPU) but do
need OS isolation, different OS versions, different OS types, or a unique version of the
application stack.
• Good choice for non-cell-based systems that need a partitioning solution (if they are not I/O
intensive).
• Applications with spiky workloads can often get more than their entitlement of CPU cycles if
the other virtual machines are not demanding those cycles.
• Easy for deploying a new application or a new instance of an existing application by
creating a new virtual machine.
Why choose Resource Partitions or Secure Resource Partitions (SRPs)?
Key Benefits
• HP Process Resource Manager (PRM) product can be used to manage system resources (CPU,
memory, and disk I/O bandwidth) according to a user-defined priority by placing processes
in processor sets (PSETs) or Fair Share Scheduler (fss) groups.
• The granularity of resource allocation for PSETs is at the whole-CPU or core level.
• The granularity for resource allocation when using fss groups is sub-CPU (as little as 1%)
• Does not require a separate instance of the OS, as do vPars or Integrity VMs.
• Memory and I/O can be shared; memory entitlements can be reallocated on line.
• Supported on both HP 9000 and HP Integrity server systems; runs on both cell-based and
non-cell-based systems.
• Can save significant amount of money on software licenses when application stacking by
reducing the number of OS instances required.
• Workload Manager (WLM) can be used to add goal-based workload management and
automation of iCAP resource usage.
• By using the Security Containment feature of HP-UX, you can place one or more secure
compartments in PRM groups to create a Secure Resource Partition (SRP). Processes in each
SRP are isolated and cannot communicate with or access the resources of processes in other
SRPs.
Trade-offs
• No hardware isolation (same as vPars and Integrity VMs).
• No OS isolation or flexibility because Resource Partitions or SRPs are in the same OS.
• Supported only on HP-UX.
• Requires same patch levels and kernel tunables because all partitions in the same OS.
8
Sweet Spots
• Excellent choice for application stacking. Can run multiple instances of the same application
on the same OS while maintaining application isolation and resource-level guarantees.
• Allows for resource-level control and application isolation without requiring a separate
instance of the OS.
HP Utility Pricing Solutions:
Why choose Instant Capacity (iCAP)?
Key Benefits
• Allows you to defer the cost of system components (processors, memory, cell boards) until you
need the capacity.
• Simplifies capacity planning; reduces the need to overprovision.
• nPars can be flexed with iCAP cores on the same server or across servers with Global Instant
Capacity (GiCAP).
• Cell-board iCAP allows you to activate a complete cell. Starting with HP-UX 11i v3, you can
dynamically add a cell to an nPar without rebooting the OS.
Trade-offs
• Available only on cell-based systems.
• To activate the usage rights of an iCAP core for a partition, the core must physically be in the
same nPar.
• Prior to HP-UX 11i v3, rebooting is required to activate an iCAP cell board.
Sweet Spots
• Provides a cushion or safety net for capacity planning.
• Provides inexpensive spare capacity that can be activated online.
• Allows you to flex the size of partitions in a failover situation.
• Provides the ability to flex nPars.
Why choose Temporary Instant Capacity (TiCAP)?
Key Benefits
• Allows iCAP cores to be activated for limited periods of time.
• When purchasing TiCAP (in 30-day increments), you can activate one or more iCAP cores
against the balance (measured and tracked in 30-minute intervals).
Trade-offs
• Available only on cell-based systems.
Sweet Spots
• Excellent choice for responding to short-term spikes in the workload; preferable to use with
WLM or gWLM.
• Cost-efficient way to provide high-capacity test environments for short periods of time.
• Allows you to lower the cost of a failover server because the cores need to be activated only
during the failover period.
9
Why choose Global Instant Capacity (GiCAP)
Key Benefits
• Allows iCAP resources to be shared across multiple servers (usage rights are deactivated on
one server and activated on another).
• TiCAP resources can also be pooled and shared across multiple servers.
• GiCAP is also integrated with gWLM.
Trade-offs
• Available only on cell-based systems.
Sweet Spots
• Can use iCAP resources for load balancing across servers.
• Can be useful in failover situations. (To flex the resources of the failover server, GiCAP usage
rights can be transferred to the failover server so that TiCAP resources don’t need to be
consumed.).
Automation Solutions:
Why choose Global Workload Manager (gWLM) or Workload Manager (WLM)?
Key Benefits
• Allows control over how shared resources are allocated between workloads.
• Automates the reallocation of resources between partitions that can share resources.
• Automates and controls the use of utility pricing to manage costs.
• Automates the activation and deactivation of TiCAP resources, ensuring they are active only
when the load requires them (resulting in lower costs).
• Allows application performance to remain at consistent levels as load varies.
• A single gWLM policy can be applied to many servers.
Trade-offs
• Both WLM and gWLM support HP-UX. In addition, gWLM supports Windows, Linux, and
OpenVMS if they are running as a guest OS on an HP Integrity Virtual Machine. Also, gWLM
is HP’s strategic workload management product.
• WLM supports Resource Partitions and Secure Resource Partitions; gWLM does not.
• gWLM may be used with Integrity VMs; WLM does not support Integrity VMs.
• WLM must be configured on each server.
• gWLM can manage multiple servers from a single management server.
• gWLM is integrated with GiCAP, providing automation and control of resources that can be
shared across servers within the same GiCAP group; even if they are in different
geographies.
Sweet Spots
• Automates sharing of processors between vPars.
• Automates flexing of nPars with iCAP processors.
• Minimizes cost of TiCAP solutions, and automatically deactivates TiCAP processors when they
are not required.
10
• Automates sharing of GiCAP and TiCAP resources across GiCAP groups
• Automates reallocation of resources in a failover situation.
11
Chapter 4: HP VSE Reference Architectures
HP VSE Reference Architectures (RAs) are documented best practices for solutions based on the VSE
components and key industry applications. The VSE RAs may provide blueprints or guidelines for
doing just what you are intending to do. HP VSE RAs are based on proven, real-world IT
deployments, and might help reduce the time it takes you to implement similar solutions. They can
also help you optimize your design time by providing examples of proven designs that you can apply
or customize to fit your specific requirements. Some of the VSE RAs that currently exist are:
• Shared IT: Shared Database Infrastructure and Shared Application Server Infrastructure
• Databases: Oracle® , and Oracle RAC
• Enterprise resource planning applications: SAP R/3 and mySAP Business Suite
• Application Servers: BEA Weblogic Server, Oracle Application Server, Oracle GRID
and IBM Websphere
• Business Intelligence: SAP Enterprise BI Server
Note:
For information about how to access the HP VSE RAs, see the last page of
this paper.
12
Chapter 5: Identifying a Pilot Project
As stated earlier, moving to a virtualized environment is a journey. It’s completely up to you whether
to start with a project that is highly visible and very important to your business, or with a smaller
project just to get oriented to a virtualized environment. HP has worked with customers who have
done both. Starting small allows you to always add new projects and new functionality later because
these technologies are designed to work well together.
Do you want to address some of your short term problems first? Is there some problem that can be
solved very simply and quickly? Do you want to replace legacy hardware or implement a completely
new project? Do you want to create an application service or utility so that new applications can be
deployed more quickly in the future?
Regardless of what project you pick, here are a just a few things to consider:
• What version of the operating system and application software will you need? If you are
moving to a new hardware platform or a new version of the OS, HP recommends that you do
a complete software-stack assessment to ensure that all of the software you need is available
and supported.
• It is wise to get buy-in from your application software providers about your new strategy or
architecture.
• Be sure that you understand the performance requirements for sizing partitions or for
application stacking. (The next chapter addresses this point.)
• Be sure to obtain management support within your own organization or from the line of
business that owns the application.
General guidelines
Here are some general guidelines and thoughts to keep in mind regarding the technology choices that
you must make for your virtualized environment.
As an alternative to having a separate server for each application, you can divide a larger server into
partitions. Partitioning solutions are great for isolating applications that have security or availability
concerns. They are also useful if applications need a different version of the OS, different kernel
tunables, or different versions of the application software. Finally, creating a new partition on
existing resources is much easier, quicker, and cost effective than provisioning a new server.
Partitioning is a relatively low-risk option because it allows you to maintain application isolation even
though the application may not be on its own physical server. In addition, if there is a concern about
availability due to multiple partitions on one server, nPars or the HP Serviceguard product might be a
good solution.
Avoid sizing for unexpected growth or for peak processing. Instead, use iCAP or TiCAP. Partitioning
solutions also allow you “right size” and move cores between partitions for handling peak loads.
Although this is not risky, the solution can become slightly more complex when using TiCAP, since you
will most likely want to add automation with WLM or gWLM to manage its usage effectively.
Application stacking with Resource Partitions or Secure Resource Partitions can be an excellent choice
for applications that work well together and that can coexist on the same version of the OS. This
solution can increase server utilization and, in many cases, can save a significant amount of money in
software licenses and support costs.
13
Chapter 6: Making Your Choices
This chapter provides an ordered set of steps that you can follow to help select your particular
implementation of HP virtualization technologies. The following steps consider the possibilities
incrementally, from those that are least disruptive to your current environment to those that represent
the greatest change and, therefore, the greatest benefit.
Determine CPU processing requirements for workloads
Before you decide on the type or types of partitioning solutions you think you want, you must first
determine the size of the Integrity server (or partition) you need for each application – just as though
you were doing one-for-one server replacement. Specifically, determine how many CPU cores are
needed on the new Integrity server to provide the performance equivalent to that of your older server.
This exercise is no different than any other server upgrade. Also determine how much memory and
what type of I/O resources are needed (for example, what kind of I/O cards and how many of each
type). In addition, identify the I/O characteristics in order to determine later in the process whether
Integrity VMs are a viable option.
At this point in the process, do not concern yourself with resource sharing. This is an additional step
in refining the solution and will be considered later. If possible, determine the base level of resource
usage for “normal” processing for your environment (whether that’s 50% or 80% utilization or
whatever number is right for your organization). Be sure you understand the resources necessary for
peak load processing. Finally, factor in a growth forecast for 3 to 5 years.
Table 1 and Table 2 provide worksheets that can help you organize your resource usage and
workload sizing information, respectively. Fill in the existing resource usage information for each
server or application workload in Table 1. After you determine what type of Integrity processors you
will be using, fill in the information in Table 2 for each server or application workload.
Table 1. Sizing worksheet: resource usage information
Legacy server
or application
workload
Number of
CPU cores
Type and
speed of
processor
Memory
required
Disk I/O
rates
Networking
I/O rates
14
Table 2. Sizing worksheet: workload information
Legacy server or
workload
Normal Processing
Peak Processing
Expected Growth
Total for Workload
Number of
Integrity CPU
cores
Memory required
Capacity Advisor
One tool that you can use for gathering performance utilization data is Capacity Advisor, a
management tool provided in the HP VSE suite of products. You can use Capacity Advisor to collect
performance utilization data from your existing servers and then to create what-if scenarios to
determine the best way to consolidate those servers. Capacity Advisor allows you to test
configurations with the size characteristics that you define, and to determine whether you can
combine various workloads onto this new server (existing or hypothetical) before you make the
changes or purchase the new server. It even allows you to factor in forecasted growth percentages
for each workload or server.
In order to use Capacity Advisor, however, you must first install the VSE Management Software on a
Central Management Server (CMS). This step might seem illogical because you are in the process of
defining your virtualized architecture. However, it might be worthwhile to install this software on a
CMS and on any managed nodes that you plan to consolidate into your virtualized infrastructure so
that you can use the capabilities of Capacity Advisor in planning your new configuration. Once
installed, you can collect performance utilization data over a period of time by using the HP
Utilization WBEM Provider collection agent for each of the managed nodes that you choose. You can
also import OpenView Performance Agent (OVPA) data if that data is available. HP also offers the
HP Consolidation Pack, which is an inexpensive limited use license for the collection of detailed server
utilization data across several key metrics. This license can be applied to the servers you are
planning to consolidate for a period of six months.
After your new virtualized environment is set up, Capacity Advisor is valuable for monitoring and
evaluating your workloads to make the most of the available systems, and for evaluating the possible
effects of moving workloads around for optimal utilization. For more information about using
Capacity Advisor, see the link at the end of this paper to the HP Integrity Essentials Capacity Advisor
User’s Guide Version A.03.00.00.
Choose the type of partitioning
The partitioning solutions are the basic building blocks or foundation of the Virtual Server
Environment. Choose one of the following partitioning options for each application or workload.
Table 3. Selection criteria for choosing type of partitioning
Type of partition Selection criteria for each application or workload.
15
nPartitions If hardware or electrical isolation required.
If work load needs dedicated hw resources.
Virtual Partitions If hardware or electrical isolation not required.
If work load needs dedicated hw resources.
If workload is not suitable for a VM: (e.g. Needs dedicated hardware
resources or is I/O-intensive.
Integrity Virtual
Machines
If hardware or electrical isolation not required.
If work load does not need dedicated hw resources.
If workload requirements are often sub-CPU (smaller than 1 core).
If workload runs with 8 cores or fewer.
If workload not I/O intensive.
Resource Partitions
or Secure Resource
Partitions
If hw/electrical isolation not required.
If applications can run on same instance of HP-UX, with same patch levels and
same kernel tunables.
Note:
If you are seriously considering Integrity Virtual Machines, refer to the white
paper entitled, “Hardware Consolidation with Integrity Virtual Machines.”
It can help you determine which workloads are good candidates for
Integrity VMs. It also provides recommendations for assessing the
performance of your current workloads (both CPU and I/O), and can help
you size the target hardware for the VM Host. There is also
another white paper entitled “HP Integrity VM Accelerated I/O (AVIO)
Overview”. See the links for both at the end of this paper.
Choose the type of Integrity server
There are many different choices for Integrity servers, ranging from server blades and entry class
server (non-cell based) to mid-range and high-end servers (cell based). Now that you understand the
hardware resource requirements of each workload, and you know which workloads are candidates
for particular types of partitioning solutions, you can decide what type of Integrity servers you need.
The information from the preceding worksheets will help you later in the process when you determine
how to combine or group partitions on a server. Additionally, it will help you determine whether or
not resources can be shared effectively between those partitions or certain application workloads.
If you need nPars or vPars, you must move to a cell-based server. If you think you will want to take
advantage of iCAP or TiCAP resources, then you must also choose a cell-based server. If you only
need Integrity Virtual Machines or want to do application stacking with Secure Resource Partitions,
you do not need a cell-based server.
If you want to consolidate applications from a lot of servers, you might want to consider the larger
cell-based servers, which will give you all of the options for partitioning. It might cost more initially
than a one-for-one server-replacement strategy, but when you consider the resource- sharing potential
as well as the ability to more rapidly create new partitions and deploy new instances of applications,
this might be more cost effective than it appears. This approach might also offer indirect cost savings
in reduced floor space, power, and cooling requirements.
At this point, you have essentially completed a server consolidation exercise and have created a plan
to reduce the number of physical servers in your data center with the addition of new servers and the
16
use of partitioning solutions. If you chose cell-based system, the next step is to consider adding iCAP
or TiCAP resources.
Utility Pricing
It’s easy to see that you should use iCAP resources if you are planning to run on a cell-based system.
Perhaps you even chose a cell-based system specifically because of the iCAP capabilities. iCAP is a
very cost- effective way to have spare capacity available for growth, whether expected or
unexpected. Most of the cost for this additional capacity is deferred until you need it, and it can be
activated dynamically with no disruption to your users. It’s a very low-risk and simple option.
TiCAP resources are useful if the increases in your workload demand are spiky or cyclical (seasonal,
monthly, or weekly). Again, this is not risky to implement, and it will add only a small amount of
complexity in order to automate TiCAP resource management with WLM or gWLM so you can better
manage its usefulness and cost.
Data from the worksheet in Table 2 can help you determine how many iCAP and TiCAP resources are
appropriate for each independent workload. The number of cores required for peak processing is the
number of TiCAP cores that you need, and the number of cores for expected growth is the number of
iCAP cores that you need. Again, at this point in the process, this data is still independent of resource
sharing. That topic is discussed in the next section.
Organizing partitions and resource sharing
Now that you understand the resource requirements for each application or workload, and now that
you have chosen which type of partitioning technology is the best fit, you must determine the best way
to organize and combine the partitions. This can get a little complicated because of the numerous
variables that factor into this decision. The first consideration is based on the sizing rules and limits.
Table 4 describes the rules and limits for each type of partition.
Table 4. Partition sizing rules and limits
Type of partition Sizing rules
nPars Must be one or more cells.
vPars One or more vPars can be in the same nPar.
Maximum number of cells per nPar is 8 (when vPars are used).
Maximum number of vPars per nPar is 8.
Maximum number of vPars in one nPar is also a factor of CPU, memory, and
I/O needed for each instance of HP-UX (cannot exceed resources of the nPar).
Integrity VMs and vPars cannot be in the same nPar.
Integrity VM Each virtual machine can have a maximum of 4 virtual CPUs.
Maximum number of virtual machines per physical processor is 20 (due to 5%
entitlement granularity).
One VM Host per nPar or server (non-cell based) is allowed.
Maximum number of virtual machines per VM Host is a primarily a factor of the
CPU, memory, and I/O requirements of each guest OS (or 254, whichever
comes first).
Integrity VMs and vPars cannot be in the same nPar.
17
The second thing to consider when organizing the partitions is how the resources might be shared.
To do this, you must further analyze each workload to understand its resource usage during peak
processing periods as well as any expected growth. Try to determine when the peak usage periods
occur and how predictable this behavior is. If possible, combine virtual machines or vPars so that
their peak processing periods don’t overlap. (Capacity Advisor is very good at helping with this
task.)
With Integrity VMs, the VM Host allocates the resources based on demand, so you can significantly
increase the utilization of a server when peak processing periods don’t overlap. With vPars, cores
within the same nPar can be dynamically moved from one vPar to another so that resources can be
applied where they are needed most. Memory can also be moved, but you would not want to do this
to address short term peaks. A good example of when moving memory resources might be useful is if
you had an application in one vPar that was busy during the day, and another application in a
separate vPar that ran at night.
If your vPars (or Integrity VMs) are combined in such a way that you have room for extra processor
cores in the nPar, then you can have a pool of iCAP cores with TiCAP usage rights. This is a very
good way to flex the size of these partitions to respond to short-term processing spikes. For vPars,
TiCAP cores can be activated for a specific vPar. For Integrity VMs, TiCAP resources are activated for
the VM Host, and then you can increase the entitlement for a specific VM. If you choose to use
TiCAP, HP recommends that you automate their activation and deactivation with WLM or gWLM.
If you have chosen nPars and need to flex their capacity from time to time, you can move the usage
rights of an iCAP core from one nPar to another by deactivating an iCAP core in one nPar and
activating an iCAP core in another. You can do this manually if the loading is very predictable and
doesn’t change too often; or you can add automation with WLM or gWLM.
Automation
To achieve the most effective benefits of resource sharing or TiCAP resources, you need to use either
WLM or gWLM. If you aren’t currently using WLM, consider gWLM for its ease of use and its ability
to define resource sharing policies for multiple systems. gWLM allows you to combine multiple
workloads with differing demand patterns on a single server to make use of idle capacity, while
guaranteeing that each workload receives its specified entitlement of resources. For multiple
workloads that span partitions or servers, gWLM can move the resources to where they are needed
most as defined by the policies you set up. TiCAP resources may also be enabled when required to
meet the workload’s demand and always disabled when not required which saves money. An
operator or administrator would simply not be able to respond quick enough to shift resources around
to meet short term demand. Additionally, TiCAP resources can be enabled or disabled at the
workload level. The gWLM tool is not as complex as it might sound. For more information about
using gWLM, see the link to the HP Integrity Essentials Global Workload Manager User’s Guide
Version A.03.00.00 at the end of this paper.
High Availability and Disaster Tolerance
It is beyond the scope of this paper to discuss HP Serviceguard high-availability and disaster-tolerant
solutions in detail. However, you might want to consider adding high availability or disaster
tolerance to your virtualized environment as you are planning your virtualized infrastructure. HP
Serviceguard is well integrated with the virtualization products. For example, in a failover situation,
gWLM can have a policy to adjust the workload entitlements on the failover server to provide
appropriate levels of resource usage to the existing workloads, along with the workload from the
failed system. Additionally, iCAP, TiCAP, or GiCAP resources can be activated.
18
Implement the Pilot Project
Now that you have determined which virtualization technologies to use for your pilot project, you
need to set up and test the new environment. As stated earlier, if your pilot project involves an
application from an ISV, be sure to let the ISV know your intentions. They might already have done
something similar and therefore might have some useful advice to help you.
One word of caution: If you are changing hardware platforms, operating system versions, or
application versions, test the transition or upgrade before you begin to install and test the
virtualization technologies. Changing too many variables at the same time is never a good idea.
Repeat the Process
Once you have successfully implemented your pilot project, you can add new functionality to it if you
need to, or you can start another pilot project. HP designed all of the virtualization technologies to
be integrated and to work well together.
19
For more information
For more information about HP VSE and virtualization technologies, see the VSE website at
www.hp.com/go/vse
White paper: “An Introduction to the HP Virtual Server Environment,” located at
http://docs.hp.com/en/11011/IntroToVSE.pdf
Virtual Server Environment Reference Architectures: follow the appropriate links at
www.hp.com/go/vsera
Book: The HP Virtual Server Environment: Making the Adaptive Enterprise Vision a Reality in your
Data Center, by Dan Herington and Bryan Jacquot, 0-13-185522-0
http://vig.pearsoned.com/store/product/1,1207,store-1974_isbn-0131855220,00.html
White paper: “HW Consolidation with Integrity Virtual Machines” located at
http://docs.hp.com/en/9984/Consolidation2.3.pdf
White paper: “HP Integrity VM Accelerated Virtual I/O (AVIO) Overview” located at
http://docs.hp.com/en/12567/avio_solution_wp-final.pdf
Technical manual: HP Integrity Essentials Capacity Advisor User’s Guide Version A.03.00.00,
located at
http://docs.hp.com/en/T2784-90026/T2784-90026.pdf
Technical manual: HP Integrity Essentials Global Workload Manager User’s Guide Version
A.03.00.00, located at
http://docs.hp.com/en/T2762-90057/T2762-90057.pdf
© Copyright 2009 Hewlett-Packard Development Company, L.P. The information
contained herein is subject to change without notice. The only warranties for HP
products and services are set forth in the express warranty statements accompanying
such products and services. Nothing herein should be construed as constituting an
additional warranty. HP shall not be liable for technical or editorial errors or
omissions contained herein.
Intel and Itanium are registered trademarks of Intel Corporation in the U.S. and other
countries. Oracle is a registered trademark of Oracle Corporation and/or its
affiliates. Microsoft and Windows are U.S. registered trademarks of Microsoft
Corporation.
4AA1-5746ENW Rev. 3, March 2009
Loading...