Dell 7000 User Manual

Stacking Dell PowerConnect
7000 Series Switches

A Dell Techni cal White Paper

www.dell.com support.dell.com
THIS WHITE PAPER IS FOR INFORMATIONAL PURPOSES ONLY, AND MAY CONTAIN TYPOGRAPHICAL ERRORS AND TECHNICAL INACCURACIES. THE CONTENT IS PROVIDED AS IS, WITHOUT EXPRESS OR IMPLIED WARRANTIES OF ANY KIND.
© 2011 Dell Inc. All rights reserved. Reproduction of this material in any manner whatsoever without the express written permission of Dell Inc. is strictly forbidden. For more information, contact Dell.
Dell, the DELL logo, and the DELL badge, PowerConnect, and OpenManage are trademarks of Dell Inc. Other trademarks and trade names may be used in this document to refer to either the entities claiming the marks and names or their products. Dell Inc. disclaims any proprietary interest in trademarks and trade names other than its own.
PC7024, PC7048, PC7024P, PC7048P, PC7024F, PC7048R, PC7048R-RA, PCM6348
July 2011
Page ii
Contents
Introduction ................................................................................................................ 2
Applicability ............................................................................................................. 2
Stacking and Management ............................................................................................ 2
Stacking and Performance ............................................................................................ 3
Stacking and Redundancy ............................................................................................. 3
Mixing M6348 and PC7000 Series Switches in a Stack ............................................................ 5
Power-Up Sequencing Considerations ............................................................................... 5
Initial Installation and Power-up of a Stack .......................................................................... 7
Selecting the Master Unit ................................................................................................ 8
Selecting the Standby Unit .............................................................................................. 8
Updating the Firmware on a Stack ..................................................................................... 9
Automatic Update ...................................................................................................... 9
Manual Update ........................................................................................................ 10
Creating a Separate VLAN for File Downloads ................................................................... 11
Adding a Stack Member with Minimal Interruption ................................................................ 16
Removing a Stack Member with Minimal Interruption ............................................................ 17
Merging Two Operational Stacks ...................................................................................... 18
Synchronizing the Running Configuration between the Master and Standby Units .......................... 18
Master Failover .......................................................................................................... 18
Effect of Master Failover on PoE Devices......................................................................... 18
Stack Member Failover ................................................................................................. 19
Failover Scenarios ....................................................................................................... 19
Scenario 1 .............................................................................................................. 19
Scenario 2 .............................................................................................................. 20
Scenario 3 .............................................................................................................. 21
Nonstop Forwarding ..................................................................................................... 22
Initiating a Warm Failover of the Manager Unit ................................................................. 23
Nonstop Forwarding Scenario ....................................................................................... 23
NSF Scenario Configuration via CLI ............................................................................. 25
NSF Example 1 ...................................................................................................... 29
NSF Example 2: ..................................................................................................... 29
NSF Re-convergence Timing ........................................................................................... 30
Medium Configuration .................................................................................................. 30
Small Configuration ..................................................................................................... 31
Stacking CLI Commands ................................................................................................ 31
Stacking Web Interface ................................................................................................. 32
Summary .................................................................................................................. 32
Page 1
Part Number
Gigabit Ethernet
Gigabit Ethernet
Modules
Other
PCM6348
161 2 4
PC7024
24
42 2
PC7024P
24
42 2 PoE4

Introduction

This white paper explains the purpose and operation of the stacking feature in the Dell™ PowerConnect™ 7000 Series Gigabit Ethernet switches. The PowerConnect 7000 series is Dell’s most advanced switching product line, offering advanced switching capabilities including high-density, high­performance stacking, and 10 Gigabit Ethernet capabilities that scale from the small business to the Enterprise Edge. Stacking allows multiple switching units to be combined together to act as a single, high-performance, highly resilient switching unit with a single management interface. Units can be added to increase throughput as needed. With each stack unit supporting up to 184 Gbps in switch capacity, the customer can have almost 2 terabits of capacity in a single stack.
In addition, stacks can be composed of an interchangeable mix of PowerConnect 7000 Series switches and PowerConnect M6348 switches, enabling administrators to continue to leverage the full utility of previous-generation switches while transitioning to the latest-genera ti o n eq uipm ent. See “Mixing M6348 and PC7000 Series Switches in a Stack” on page 5.

Applicability

This paper applies to the PowerConnect 7000 series switches, which includes the PC7024, PC7048, PC7024P, PC7048P, PC7024F, PC7048R, and PC7048R-RA Dell model numbers. Each PowerConnect switch has two bays that can be customized to support a stacking or an uplink configuration. Each bay can contain a CX-4, SFP+, or a 10GBase-T module. Stacking is supported only on CX-4 modules in either or both bays. CX-4 stacking modules can be configured in 16 Gbps stacking mode or 10 Gbps Ethernet uplink mode. CX-4 stacking modules default to stacking mode, where the maximum cable length is 3 meters. When the stacking modules are configured to operate in Ethernet mode, the maximum cable length is 12 meters.
The PCM6348 has two 10 Gbps SFP+ ports, and two separate 10 Gbps Ethernet ports that support CX-4 modules for stacking.
The following table summarizes the features of the PowerConnect 7000 series and PCM6348 switches.
Ports (RJ-45)
PC7024F 4 24 2 PC7048 48 43 2 PC7048P 48 43 2 PoE4 PC7048R/ PC7048R-RA 48 43 2 Top-of-Rack5
1. The PCM6348 also has 32 in ternal ports that conn ect to server blades in a chassis.
2. Shared with ports 21–24.
3. Shared with ports 45–48.
4. Each copper port can provide up to 30W of power to an external powered device.
5. The difference bet wee n the PC7048R and PC7048R-RA switche s is t he airflow direction.
Ports (SFP)
(Stacking, SFP+, 10GBASE-T)
Features

Stacking and Mana gem e n t

An important advantage of stacking is that it provides a consolidated interface for management of multiple switches when linked together. When a stack is already deployed in the network, operators can add units to the stack as their port requirements increase, with minimal administrative overhead
Page 2
Recommended
Not recommended
Module
A
Switch
1
2
3
4
5
Switch
1
2
3
4
5
Module
A
aCables connected in a complete ring.
r
Cables not connected in a complete ring.
required for reconfiguration. Additional stack members can immediately utilize existing configuration information such as routing and switching configurations, VLANs, ACLs, port profiles, and security certificates.

Stacking and Perfor m a n ce

For situations where there is a need to pass traffic between switches and the aggregate bandwidth required between PowerConnect 7000 Series switches does not exceed 64 Gbps (2 ports, 16 Gbps Tx
and Rx each), a stacking configuration offers an attractive alternative to Link Aggregation Groups
(LAGs). Stacking configuration is generally transparent to the operator and does not require configuration beyond cabling. In addition, failover times are generally faster in a stack configuration than in spanning tree or link aggregation group configurations. Note that other PowerConnect Series switches may have different supported bandwidths for stacking.

Stacking and Redun d ancy

By connecting a cable from the last switch in a stack back to the first switch, the operator ensures that a stack has the protection of redundant paths for control and data traffic, including support for LAGs configured across multiple switches. This means that any single point of failure (a switch or a stack cable failure) will not affect the overall operation of the remaining stack elements.
#Figure 1 shows two stacking configurations with a single stacking module on each switch. In the recommended configuration, a stacking cable connects each switch to the next, and the top switch is connected to the bottom switch to form a complete ring.
Figure 1. Stack Configured in a Ring
Page 3
Recommended
Not recommended
ModuleAModule
B
Switch
1
2
3
4
5
Switch
1
2
3
4
5
ModuleAModule
B
a
Cables connected in a ring.
a
Dual stacking cables connect both ports
r
Cables not connected in a complete ring.
r
Dual stacking cables connect ports on a module to two different switches.
on a module to the same switch.
#Figure 2 shows a recommended stacking configuration with dual stacking modules on each switch. Each
switch connects to another with 128 Gbps of total bandwidth (4 ports @ 16 Gbps Tx and Rx each). Note that in the recommended configuration, the dual stacking cables from each module both connect to the same switch. In the not-recommended configuration, the dual stacking cables on switch 2, module B and switch 3, module B connect to two different switches.
NOTE: To ensure full bandwidth when using redundant links between two switches, be sure not to split the links across modules. Keep redundant links isolated to a single module on each end as shown in the recommended configuration in #Figure 2.
Figure 2. Dual Stacking Modules Configured in a Ring
Page 4
PC7048
PC7048P
PCM6348
PCM6348
PC7024
1
2
3
4
5

Mixing M6348 and PC7000 Series Switches in a St ack

PC7000 series and M6348 can be used interchangeably in a stack of up to 12 units. As with PC7000­series only stacks, all switches in a mixed stack must have the same firmware version. Starting with the
4.1.0.6 software release, a single image supports both switches. No additional configuration is required.
Either the PC7000 series switch or the M6348 switch can be the master. NOTE: A stack that includes an M6348 switch (as the master or a member) can operate only in Normal
mode—not in Simple mode. #Figure 1 shows a stack with a mix of PC7000 series and M6348 switches.
Figure 3. Mixed Switches in a Stack

Power-Up Sequencing Considera t ion s

One unit in the stack acts as the master unit. The master manages all the units in the stack. A second switch is elected as the standby u ni t, w hic h be co mes t he master if the master unit is unavailable.
The administrator can manually configure which unit is elected as the standby, or the system can select the standby automatically. To configure it manually, the administrator can use the Stacking Management > Unit Configuration page in the web interface or the standby command in the CLI.
NOTE: The terms “master” and “manager/management unit” are used interchangeably throughout this document.
The selection of the manager and standby units is important in situations where the administrator wishes to utilize the serial port for management, perhaps as a backup console. Management of the stack via the out-of-band service port or the in-band ports is transparent in a stack.
In a stacking configuration, the power-up sequence determines the manager and standby units. The switch that the operator selects as the manager should be powered up first and should be allowed to fully come up before the other stack units are powered up. The standby switch should always be directly connected to the manager. Once the manager and standby have powered up fully, other
Page 5
members of the stack can be powered on sequentially by powering up the switch adjacent to the last switch powered on.
A stack of units is managed and acts as a single entity when the units are connected together and are operational. If a unit cannot detect a stacking partner, the unit automatically operates as a stack of 1 with itself as the master. If a stacking partner is detected, the switch always operates in stacking mode.
When units are operating together as a stack, the following activities occur:
All units are checked for firmware version consistency on startup. By default, units with older versions of firmware are automatically upgraded with the operational firmware version on the master switch as they join the stack. If a unit has a newer software version than the master, it is not downgraded to the master’s version by default; however, the administrator can use the following command to enable this functionality:
console#boot auto-copy-sw allow-downgrade
Before adding a member that has newer version of software to a stack, the administrator should enable the automatic downgrade feature on the master switch.
Switch management and protocols such as OSPF are active only on the master, but apply to all members of the stack. Unless administratively disabled, the Nonstop Forwarding (NSF) feature periodically checkpoints the running configuration and the application state betw ee n t he master and standby switches during normal stacking operation. If the master fails, the standby switch takes over operation of the stack.
Data forwarding is active on all units in the stack, including the master. Data forwarding continues to operate should the master become unavailable.
Page 6

Initial Installation and Power-up of a Stack

Follow these instructions to create a stack: NOTE: Install units in a rack whenever possible to prevent the units and cables from being disturbed.
1. Install all stacking cables. Fully connect all cables, including the redundant stack link.
CAUTION! We highly recommend that a redundant link be installed to provide stack resiliency.
2. Select a unit to be the manager unit. Power this unit up first.
3. Monitor the console port on the manager unit. The unit will automatically become a manager unit. If not, renumber the unit as desired. The Stack Management > Unit Configuration page in the web interface or the switch renumber command in the CLI can be used to change the unit number.
4. If desired, pre-configure other units to be added to the stack. This is not usually necessary.
5. Power on a second unit, making sure it is adjacent (directly connected) to the ma na g er unit. This ensures the second unit comes up as a member of the stack, not as the manager of a separate stack.
6. Monitor the manager unit to see that the second unit joins the stack. Use the show switch command to determine when the unit joins the stack. It will be assigned a unit number (unit #2, if it has the default configuration). The output of the show switch command should indicate the unit status as “OK” if the member has been successfully added to the stack. If the unit status is "code mismatch" and stack auto-upgrade is disabled, then use the copy image unit command to update the code on the unit. If the unit status is "Cfg Mismatch", then use the no member <unit-number> command to resolve the issue.
7. If desired, renumber this stack unit using the switch renumber command.
8. Repeat steps #4 through #6 to add additional members to the stack. Always power on a unit directly connected to the units already in the stack.
9. Enter the show switch stack-port counters command on the manager and see if there are any stack port errors being reported. Replace the stacking cables/modules if stack errors are being reported and begin the power-up sequences again for the affected units and any downstream units.
Page 7

Selecting the Master Unit

A stack manager is elected or re-elected based on the following considerations, in order:
Whether the switch that was previously the stack manager.
Whether the switch has the higher MAC address.
When a switch is added to the stack, one of the following scenarios takes place:
If the switch was previously designated as the stack master but another master unit is already active, then the switch changes its configuration to be a slave unit.
If the switch was not designated as the stack master and there is another stack master in the system, then the switch changes its configuration to be a slave unit.
If the switch is enabled as the stack master or there is no other stack master, then the switch becomes stack master.
If the switch is not enabled as the stack master, the unit remains a slave unit.
The administrator can manually set the unit number for the switch using the Stack Management > Unit Configuration page in the web interface, or the switch renumber command in the CLI. To avoid unit-number conflicts, one of the following scenarios takes place when a new member is added to the stack:
If the switch has a unit number that is already in use, then the unit that is added to the stack changes its configured unit number to the lowest unassigned unit number.
If the added switch does not have an assigned unit number, then the switch sets its configured unit number to the lowest unassigned unit number.
If the unit number is configured and there are no other devices using the unit number, then the switch starts using the configured unit number.
If the switch detects that the stack already has the maximum number of units, making it unable to assign a unit number, then the switch sets its unit number to “unassigned” and does not participate in the stack.
If a new switch is added to a stack of switches that are already powered and running and already have an elected master unit, the newly added switch becomes a stack member rather than the master. On the master unit, if there is no saved configuration for the newly added unit, it applies the default configuration. If there is a saved configuration on the master for the newly added unit, it would apply the saved configuration to the new unit. If the entire stack is powered OFF and ON again, the unit that was the master before the reboot will remain the master unit after the stack resumes operation.

Selecting the Standb y Unit

When the stack is formed, one of the units is automatically selected as the standby unit for the stack. The standby unit takes over as manager if the current manager fails. Alternatively, the administrator can specify the standby unit.
To configure the standby unit using the CLI, use the standby command in Stack Configuration Mode.
Page 8
To configure the standby unit via the Web interface, view the Stacking > Unit Configuration page and select Standby for the Unit Type.

Updating the Firmware on a Stack

Automatic Update

By default, firmware synchronization is performed automatically when a unit is powered up on a stack. Because firmware synchronization makes no attempt to check for the latest version of firmware, the following procedure is recommended for bringing all members of a stack onto the same version of code.
NOTE: Schedule some downtime as this process will reset the entire stack and affect all stack users.
1. Use the following commands to enable firmware synchronization, if needed, and write the running configuration to the saved configuration:
console#boot auto-copy-sw console#copy nvram:config copy system:running-config nvram:startup­config
2. Load the more recent firmware image onto the stack master as the backup image.
3. Set the stack master to reboot from the new image:
console#boot system backup
Page 9
Loading...
+ 23 hidden pages