8200 zl Series Redundant Management Module

[Vince Roth]

We are looking at purchasing an 8206 zl switch to replace our existing network infrastructure.

Can anybody tell me if the switch is configured with redundant management modules, can the firmware be upgraded without rebooting or otherwise taking the switch down?

[David_Schwartzs]

The 8200zl series switches can be configured with redundant management modules. If you're buying a stock 8206zl, it only comes with one module - you can add a second one by purchasing part # J9092A as an add-on.

There is a process for updating the firmware on both modules and then making the new firmware active. There is a minimal amount of downtime to switchover to the new software, but with redundant modules, you don't have to wait the full time for a full reboot because you can reboot the modules individually, allowing you to do a graceful switchover between the active module and the standby module rather than waiting for a boot cycle.

This process is detailed here:
http://h20000.www2.hp.com/bizsupport/TechSupport/Document.jsp?lang=en&cc=us&taskId=125&prodSeriesId=3437443&prodTypeId=12883&objectID=c02597106

Let me know if this helps you out.

[dmesser-hhs]

You will still have a service interruption as dual mgmt cards do not provide a "hitless" upgrade option.  However as David mentioned the boot time is minimized.  In my experience on my 8212zl's the outage is reduced by 1/2 as compared to a full  boot cycle.

Dominic Messer

[Vince Roth]

Thanks, so this wouldn't be a good solution for us to use with an iSCSI SAN then, given the criteria of not wanting to have to take the iSCSI SAN down first (as we have a lot of VM's that take a lot of time to take down and bring back up).

Can anyone suggest a "hitless" network architecture, where it is safe to perform a firmware upgrade on a switch without risking data corruption on an attached live iSCSI SAN?

[David_Schwartzs]

This is why distributed scale out iSCSI SANs are awesome - particularly the HP StoreVirtual SAN (formerly HP Lefthand P4000).

If you're using the HP P4000 SAN in a best practices configuration, you have redundant switches and redundant nodes. Using HP's Network RAID 10 to provide synchronous replication services across nodes, you essentially have two SANs that are operating as a single SAN, providing both redundancy as well as performance increases. You can also take half of a mirrored pair offline for planned maintenance like firmware & software upgrades, allowing for SAN upgrades without any downtime at all. It eliminates all single points of failure. If you're using iSCSI SAN but not using HP StoreVirtual... you should take a serious look at it.

To a lesser extent, you can do some of this with other SANs (at least basic network path redundancy) - assuming the SAN has more than one network port, they should be connected to different switches (i.e. you should always have at least 2 SAN switches in best practices). This way, if one switch goes offline, planned or unplanned, your servers always have an alternate route to the SAN.

Redundant management modules just protects from one aspect of failure (regardless of whether downtime is planned or unplanned) - there should always be redundant switches so this simply isn't an issue.