VMware (ESX 5.5) and SAN question about datastores

I think I have asked this question before, but I wanted to revisit and get some more clarity.

If I have three or four ESX hosts all connected to the same shared storage of 24TB, what is the best way to chop up the storage?

Should I make one large datastore of 24TB and attach it to all hosts, or even two smaller 10TB datastores or should I chop it up into smaller LUNS, like 2TB?

Here's what I'm seeing.  We are hosting about 600 virtual machines in a farm, spread across about 10 servers.  What I see when I log into vCenter is that the storage was chopped into several 2TB datastores and a few VMs are stored on datastore1 and a few VMs are stored on datastor2 and a few VMs stored on datastore3, etc.

So, now, I have ten 2TB datastores, each hosting about 60 VMs on each one.

My individual VMs are growing and I'm needing to expand some, but I'm limited because the space on each individual store is not enough to expand all of the machines on that store so I'm having to juggle.

Cumulatively, the total free space on the SAN is enough, but since it's all spread out all over the place, I can't utilize it.

My question to the client was "Who chopped up your SAN like that and why?"  

I would have made it one large store @ 24TB or maybe 2 @ 12TB and then put all of my VMs out there and expand them as I needed.  The way it is now, with it all chopped up, I'm having to look at sizes of VMs, free space on stores and move things around manually.  

So, the question is:  Do you chop up your array or delve it out as one big bucket of storage?


Who is Participating?
Andrew Hancock (VMware vExpert / EE MVE^2)VMware and Virtualization ConsultantCommented:
If I have three or four ESX hosts all connected to the same shared storage of 24TB, what is the best way to chop up the storage?

We break the datastores up into datatores, of sizes 500-800GB, and then some datastores are supersized for 1TB, and some that are 2TB, and 3TB for Large VMs.

We have at least 11+ datastores created on a SAN of size equal to 24TB, but we do not provision all at once, we provision as we need storage and datastores.

We do not like supersize LUNs, for performance, and also corruption, e.g. a 24TB datastore gets corrupted that's many VMs which also get corrupted.

Also we use Snapshots and Replication which can also dictate size of LUNs.
Alessandro ScafariaInfrastructure Premier Field AdministratorCommented:
This is a very "fair enough" question.....

You start point has to be this: "Storage space is never enough" :-)

Setting up LUNs depends on a lot of factors, such as performance and initial space required.....

RAID 10---->it's the current trend
RAID 1---->a good compromise in speed
RAID 5----->VMs with poor performance (DCs, fileservers....)

Sometimes you NEED a RAID 10 because your VMs needs to be faster than others and you may have 2 spindles in fault without losing anything....

Sometimes you don't need this and you accomplish a simple RAID 1.....

Sometimes you swift with RAID 5 and you will be fine too (only for poor performance VMs).....

Basically puts all eggs in the same basket is unconventional because of performance and corruptions issues (you will loose everything).....usually you need always a 2nd plan in your IT life and some differences in your SLAs with clients.....

Some best practices say that you may setup a SAN looking to RAID types.....
In this scenario, you create a RAID 10 LUNs containing (and planning a reasonable growing space plan) your faster VMS, then you may setup RAID 5 LUNs for your slow VMs and finally some RAID 1 LUNs for you intermediate VMs....

Usually if you don't know how VMs may grow up, you should have the budget to buy another SAN for your purpose.....

VMware or Veeam then may help you to easily move your VMs in a new datastore without almost any downtime.....

Waiting for Experts opinion, let me know your thoughts.....
Andrew Hancock (VMware vExpert / EE MVE^2)VMware and Virtualization ConsultantCommented:
and if you have a SAN, you may not be able to set RAID 10, RAID1 and RAID 5, because if you use Example:- A NetApp filer, all disks in the shelves, are all RAID-DP!
Question has a verified solution.

Are you are experiencing a similar issue? Get a personalized answer when you ask a related question.

Have a better answer? Share it in a comment.

All Courses

From novice to tech pro — start learning today.