ESX 3.5, IBM Bladecenter and SVC 4.2

[samarripas]

We are in the process of implementing ESX running on IBM

Bladecenters (Brocade FSM) attaching to IBM SAN Volume Controller. We keep running

into snags with brocade FSM's and fabric issues. We cant find best practice for

ESX behind SVC as well. We have been told to run 500GB LUNS and others tell us

it does not matter. Is anyone out their running a similar environment? Curious

to see how it is working for others.

Message was edited by: Texiwill - Removed weird style information

[Texiwill]

Hello,

Well it does matter actually.... You want as many spindles as you can use per volume, so if you can make one large meta LUN this would be better than individual LUNs. You then divide the meta LUN into your real LUNs to present to ESX.

I have IBM hardware but not SVC. 2TBs is the max size of a LUN however most people opt for multiple LUNs as it increases the total number of SCSI Reservations that can be made and reduces SCSI Reservation conflicts based on the load of the system.

Best regards,

Edward L. Haletky

VMware Communities User Moderator

====

Author of the book 'VMWare ESX Server in the Enterprise: Planning and Securing Virtualization Servers', Copyright 2008 Pearson Education.

SearchVMware Blog: http://itknowledgeexchange.techtarget.com/virtualization-pro/

Blue Gears Blogs - http://www.itworld.com/ and http://www.networkworld.com/community/haletky

As well as the Virtualization Wiki at http://www.astroarch.com/wiki/index.php/Virtualization

[samarripas]

We have that covered. We have configured our SVC to utilize 24 spindals for ESX LUNS/VDISKS. If you dig into how the SVC works it will shed light on how they virtualize disks with in multiple LUNs. That is why I am looking for someone that is using hte SVC with VMWare. We use 2 x DS4800's on the back end for storage. We have a lot of technology and a tone of options. We are still getting familiar with the technology.

[Jwoods]

I would agree that it does not matter. We have a similar setup using IBM HS20 Blades, DS8300/SVC 4.2 within a Cisco fabric. We had that debate as well and chose to make 300GB LUNs and 600GB LUNs. I'm not familiar with Brocades' fabric manager. Regardless of small or large LUNs we get excellent performance. We utilize multiple SVC IO groups and make sure that our ESX LUNs are balanced across those groups. One important thing is to test test test. We ran multiple IOmeter tests with different configurations to see what config performed the best. Almost each configuration test yielded results acceptable for a production environment.

[davidbarclay]

Just because you've virtualised your storage (SVC), doesn't mean the same basic principles don't apply. We have a number of customers with SVCs..when used incorrectly you can hav ethe same problems that users with entry level storage can have.

In fact, you have more considerations to be concerned with as it can add more complexity. As mentioned, test before you assume anything. Also read the best practice guides. Also talk to IBM if you aren't sure, because at the end of the day the buck will stop with them, not VMware.

Dave