Future of VMFS file system

[bstrang]

We are planning some expansion of our ESX clusters and SAN storage, and I was trying to find any information about changes/features coming up for the vmfs file system with the new release of esx(vsphere). Is there going to any changes that will require "conversion" of file system with new ESX as with going to ESX 3?

[azn2kew]

All informations are under Non-Disclosure Agreement (NDA) so if anything out there is just assumption, and wait VMware gives you the right answer. But from looking back previous version, anything that upgrade/migration would be flexible and easy for everyone. So, I'm sure VMware aware of this migration/upgrade requirements and it should be supported in some ways. I'm confidence VMware has the best solution for all of us who want to uprade to vSphere & vCenter 4.0 from all aspects of networking, storage, security standpoints.

If you found this information useful, please consider awarding points for "Correct" or "Helpful". Thanks!!!

Regards,

Stefan Nguyen

iGeek Systems Inc.

VMware, Citrix, Microsoft Consultant

[glynnd1]

You should be able to Google up the documents that cover moving from ESX 2.x to 3.0. This will cover the steps that people went through in dealing with the change from VMFS2 to VMFS3, and the upgrade in VM "hardware". While it was time consuming it was not overly bad as I recall.

[vSeanClark]

2.5 to 3.0 was a pain in the butt, but not too bad if you had plenty of host capacity to do a "migration upgrade". (Re-installing ESX on one of the servers and then restarting VMs on that newly loaded ESX server 1 by 1. ) If you would have waited til ESX 3.0.1 was release, there was a storage VMotion-type feature that allowed you to do storage VMotion to the new VMFS and then simply reboot the VM when done. I would imagine that they will have something newer and schnazzier for the 3.5 to 4.0 migration/upgrade path.

I would start making plans for a rolling migration upgrade whenever you make the jump to the next gen. That would mean monitoring capacity on your ESX servers so you can be sure you have at least n+1 capacity so an ESX server could be taken out of the cluster and reloaded without undo strain on your VM workloads.

Cheers!

Sean Clark - -

[glynnd1]

My memories must be of 3.0.1, we weren't rushing to upgrade, and yes additional host and storage capacity was useful.

[Erik_Zandboer]

Hi,

I do not know how complex it will be, if it will be even neccesary to upgrade - but when looking at 2.5 -> 3.0, which was a major upgrade on VMFS (VMFS2 to VMFS3), it is all pretty painless, as long as you have an empty LUN. You can upgrade the empty LUN without risk, then move some VMs, upgrade the second LUN, etc. especially with hot cloning being about, it should be hassle free. So do exend your storage by all means, it can make a migration (if needed) only easier.

Visit my blog at http://erikzandboer.wordpress.com

[azn2kew]

I'm assuming the person want to know vSphere future VMFS compatible with upgrade/migration process from ESX 3.5 to vSphere 4.0? If not, than you have two options in place upgrade or side by side migration, and depends how you want your hosts/guests downtime be. I would choose side by side migration and stand up new ESX 3.5 clusters with plenty of LUNs presented and migrate your hosts and vmdks and than upgrade the guests hardware and VMFS2->VMFS3. If you plan it correctly, the process is straight forward.

If you found this information useful, please consider awarding points for "Correct" or "Helpful". Thanks!!!

Regards,

Stefan Nguyen

iGeek Systems Inc.

VMware, Citrix, Microsoft Consultant

[glynnd1]

Correct Stefan he is. But anyone who knows that answer is under NDA, and therefore can not answer the question.

Hence the the topic has migrated to 2.x to 3.0 upgrade with the presumption that at it's worst it will be like that.

I think we can safely say that since then VMware has learned a thing or two and will have improved the VMFS upgrade process, that is if there is an upgrade.

[bstrang]

Thanks for all the input...

We didn't have to do the 2.5 to 3.0 upgrade so I was spared that migration. I will have one additional ESX host on each of our clusters(16 hosts on each) but extra storage may be an issue. We have 20TB now and by the time we upgrade we may be close to 30TB split up in 500GB LUN's. At most I might have 1 spare 500GB LUN(one again on each cluster). Is there a problem with LUNs at different levels in the same HA/DRS cluster?

[azn2kew]

You don't have to have exact LUNs size for side by side migration, 500GB is good enough to stage for targeted guests as much as you can and then once completed go back and wipe old LUNs and reprovision it again so that you can keep migration process. Any types of LUNs are fine as long as they all visible to both environments and the only issue is performance if you have different RAID configurations but that's your storage designs.

If you found this information useful, please consider awarding points for "Correct" or "Helpful". Thanks!!!

Regards,

Stefan Nguyen

iGeek Systems Inc.

VMware, Citrix, Microsoft Consultant

[Erik_Zandboer]

It is more complicated than that. You need to be able to free up an entire LUN before you can upgrade that LUN. So 500GB is not always enough; if you have for example 600GB LUNs which are filled up to 550GB you would not be able to perform an upgrade this way... Unless you could move one of the smaller VMs to yet another LUN... It can be quite a puzzle. But, this is all academic. Better to wait for VI4 and see!

Visit my blog at http://erikzandboer.wordpress.com