Snapshot best practices guide for Equal logic

If you're using SAN-based snapshots, you can see in the group manager how much space is being used for these snapshots (the middle one of the 3 bars when looking at the volume). If that bar never goes over, for instance, 100GB, you could reduce your snapshot reserve (done per volume) to 25% or so (20% plus a little extra just in case) and free up ~375GB.

Thanks for your time.

Thats a good pointer. I was able to find some useful info, but have few questions about it.

For eg: we have a volume say (s1-vol4) and the space utiliized in this volume is 324GB and utilization of the snapshot space is 237.22 GB

And for this volume it is configured to take one snapshot at night 12:00 AM. And it retains only one snapshot at a point in time.

Questions:

1. How the snapshot utilization is less as compared to the volume utilization.?

2. Is there any recommendation that we should have minimum two/one snapshots to be retained.?

Iam new to this , so it would be great if you could provide some guidelines on how to configure the snapshot space and make best use of the volume space.

From the above example stated , I understand that if i want to have one snapshot per day , i can reduce the snapshot reserve percentage., but iam not sure about the best practices that can be done.

Also in the above example i have mentioned the volume utilization is 324GB , but that value i got from the EQL logic browser , but the same volume size when i find using the ESX host, it shows the value differently as below

[root@ombb-vm12 ombb-eqlx-s1-v4]# vdf -h .
Filesystem            Size  Used Avail Use% Mounted on
/vmfs/volumes/4ca9602b-bbc2926a-8e6f-0023aef90340
                      499G  409G   90G  81% /vmfs/volumes/ombb-eqlx-s1-v4
[root@ombb-vm12 ombb-eqlx-s1-v4]#

Prabs.

Prabs,

Hi, I'm Joe with Dell EqualLogic.

Q1: Snapshots are just used to hold the data that has changed on the base volume, and the data that may have changed on a snapshot you've mounted, since the snapshot was created. So the instant you create a new snapshot, it uses zero space. If your volume is absolutely never used, you could have hundreds of snapshots (up to the maximum of 512) that occupy together no space at all.

As your volume is used, though, your snapshot grows. Every time you write to a disk page that you haven't written to since you created the snapshot, the snapshot is given the old data, and the new data is written to the base volume. This allows us to make an exact representation of the volume at the time the snapshot was created, with a minimum of actual extra storage space. But any server that mounts a snapshot will *see* the disk as it was at the time the snap was made, even though we're only storing the difference from then to now.

So a snapshot grows as long as a volume is being used. And any writes you make to a snapshot if you login to it directly will also cause a snapshot to grow, since we have to store that data separately from the base volume. No write to a snapshot will ever get back to your base volume

Q2: That is a business decision that you would need to make.  This should be based on your business continuity/disaster recovery plans, disk space available for snapshots, and your backup strategy.  However with that said, a general guide line is to ask the question “how far back in time, would be acceptable to have to reenter data, if a problem arises”.  So in your case if you had a failure at 1PM the following day, and you needed to restore the snapshot, you would lose all the work created on the volume from 12AM on (a total of 13 hours).   Generally speaking, most would choose something in the 4-8 hour range for non-critical, and 1-2 hours in the super high critical type situations.

Regarding you last question where the host see's different sizes then the array.  Please review this post by dwilliam62 which gives the best answer: http://en.community.dell.com/support-forums/storage/f/3775/t/19368414.aspx

Regards,

Joe

Thanks for the detailed answer Joe.Iam with DELL PG.

I got the answer what i was looking for.

My plan is to have all critical Guest machines (VM's) in one or two LUNS and create a snapshot reserve according to the use and I will determine on the time interval to run the snapshot schedule.

But for other non-critical VM's I feel either there is no need of a snapshot of the volumes in which it reside or i will keep the snapshot reserve small and run the schedule once in a day.

Once again thanks for the information.

Incase of some issues with the volume , is there any recovery from snapshot procedure available, that would be great if you send it across.

Prabs.

If you are going to be snapshotting volumes that contain VMs you should use AutoSnapShot Manager Vmware Edition. One of the main benefits is that you can restore individual VMs easy.  It allows you to create scedules also and it can quiesce the operating system of the vms also.  I think if you snapshot the volumes only by using the snapshots that you can do using the array the vms may not work if you attempt to restore them.