Unsolved
This post is more than 5 years old
9 Posts
0
831
Implementing VP Snap on VMAX - separate pool vs. dedicated pool
Hello,
Trying to implement VP snap in my environment and getting stuck on a decision point.
My vmax currntly have three vp pools defined (one for FC, one SATA and one for EFD).
This vmax is currently maxed out, I cannot add any additional capacity to it.
I am trying to figure out if I should drain some of my existing datadev from each of my existing vp pools and use them to create three new vp pools for vp snap.
Reason why I am thinking of three new dedicated pools for VP snap is to eliminate pool overrun and ultimately bringing my production down to a scretching halt. We all know these things happen, DBAs can essenciately turn these snaps in to regular full clone and hurt not just themselves but also others who are using these pools.
Question is, does it make sense to create separate pools?
Fenglin1
2.1K Posts
0
August 7th, 2013 22:00
BTW, your question written to the users' own "Discussions" space don't get the same amount of attention and questions can go unanswered for a long time, You can select "Move" under ACTIONS along the upper-right. Then search for and select: "Symmetrix Support Forum" which would be the most relevant for this question.
Fenglin1
2.1K Posts
0
August 7th, 2013 22:00
Creating dedicated pools for VP SNAP an alternative. But I don't think there any difference between you using existing pools on performance side. As the tracks being moved by host write I/O still resides on you data devices on your physical disk. From operation perspective, as you said your VMAX is maxed out, draining space and creating will be much complicated. As i do , I will select creating new Thin Device and bind to existing pool, then create VP SNAP and leaving VMAX to do the rest.
Just for you reference.
xubair1
9 Posts
0
August 9th, 2013 07:00
Li, thanks for the feedback.
Yes, agreed, from a performance standpoint, it doesn't make too much sense to have a separate pool.
I was more concerned from a capacity standpoint. I don't want my test/development hosts (who happen to be using these VP snaps), filll up my existing pool with their write requests and put my prodution hosts in jeopardy for writes.
To avoid situation like this, I was thinking of segragating/confining my test/development hosts into their own pool.
Am I being a little paranoid here?
And extra thanks for the 'move' suggestion which i have done already
Fenglin1
2.1K Posts
0
August 11th, 2013 19:00
Yes, I agree. If your concern is that the test/dev hosts are going access much extents in your production VP pool. But i believe it will also occurs, when you use separate pool, your test/dev hosts access original extents in production VP pool. In that case, I think you'd better to create TF/Clone than TF/Snap. But you do have capacity challenges, I will stay on my first reply and create new thin device and bind to existing pool.
That's my advice for your reference.