Comments on: ESX4 ALUA, TPGS and HP CA

By: VCAP-DCA Study Notes – 1.3 Complex Multipathing and PSA plugins | www.vExperienced.co.uk

Sat, 16 Apr 2011 09:55:25 +0000

[...] Determining whether an array is ‘true’ active/active is not as simple as you might think! Read Frank Denneman’s excellent blogpost on the subject. Our Netapp 3000 series arrays are asymmetric active/active rather than ‘true’ [...]

By: Aloha ALUA | DeinosCloud

Aloha ALUA | DeinosCloud — Sun, 03 Apr 2011 13:04:59 +0000

[...] that ALUA exactly?, Pluggable Storage Architecture, exploring the next version of ESX/vCenter, ESX4 ALUA, TPGS AND HP CA, Understanding more about NMP RR and iooperationslimit=1, See a list of Storage Array Type [...]

By: Saturnous

Saturnous — Tue, 30 Nov 2010 08:04:15 +0000

I missing the important hint that “Implicit Lun Transfer (ILT)” is disabled for LUNs in DR groups. In VI3 this was a huge pain point because you set the pathing equal over the cluster and relied that the EVA will switch. It never switched over the DR group and the EVA suffered. Worst when using only one DR group having all LUNs on one Controller only. With vSphere where you determine the ownership from EVA side this is not a issue, but another come up – resignaturing.

Anyone has an idea how to run CA over 2 different EVA Arrays with vSphere ?

ESX 3.x you set DisallowSnapshotLun = 0 and HA + Kernel will take care.

In vSphere Snapshot handling is different, but what will happend when the same NAA came up with different model ? It will need to remount it and mask the LUN.

Or does the ESX remember a forced mount forever (this DS can show up with this and this parameters) . I assume a forced mount put the LUN properties into the VMFS header (need to hexview diff this).

My “simple Campus CA EVA SAN with different EVAs non-SRM” approach i will try tomorrow is creating a Script looking for serious amount of path dead conditions – masking the “failed EVA” per device name (HSV210 / HSV300) – remounting the data-store forced. The problem is it assumes a complete EVA will fail – which is actually not a issue where only one LUN is shared. Also the approach is not easy to test in a hot environment with a single test LUN in a DR group. Masking the controller number with LUN would allow a hot test, but what happens when the Host reboots without access to the failed EVA – will the controller number remain the same ? Will it mask the remaining EVA too (in case the lower EVA will fail) ? Can i assign a fixed controller number with “persistent binding” within the HBA ?

By: Greg

Greg — Wed, 24 Nov 2010 03:15:19 +0000

Just what I needed. Thanks

By: Warehousing

Warehousing — Fri, 16 Apr 2010 22:48:07 +0000

I never, ever would have imagined I would be required to be familiar with this, but thank goodness for the internet…

By: New home: http://frankdenneman.nl « Frank Denneman

New home: http://frankdenneman.nl « Frank Denneman — Wed, 31 Mar 2010 19:03:20 +0000

[...] March: ESX4 ALUA, TPGS and HP CA 25th March: Identify storage performance issues 16th March: VCDX Tip: VMtools increases the Disk [...]

By: Mike Mills

Mike Mills — Wed, 31 Mar 2010 11:20:50 +0000

At my company we have 3 EVA 8100s and 4400. We are in the initial planning phases for our vsphere migration, and I really enjoy this (and other) articles regarding the best way to handle things between vsphere and EVAs. Thanks so much!

- Mike

By: Top 5 Planet V12n blog posts week 12 | VMvisor

Top 5 Planet V12n blog posts week 12 | VMvisor — Sun, 28 Mar 2010 23:31:54 +0000

[...] Denneman – ESX4 ALUA, TPGS and HP CAVMware addressed this shortcoming and added ALUA support in the new storage stack of ESX4. The ALUA [...]

By: Raiko Mesterheide

Raiko Mesterheide — Fri, 26 Mar 2010 19:35:22 +0000

Hi Frank,

very great article. Especially the AAA and SAA differentiation wasnt that clear before reading!

Good job!
Raiko