1.3K Posts

June 26th, 2013 07:00

If this is a VP environment, then the size of the host volume (TDEV) has nothing to do with the size of the hyper on the disk.

There is usually some relationship to capacity and IOPs, sometimes called access density.   So with more capacity comes more IOPs.

However if you have the same workload and the same data and the same number of meta members, the performance will be the same, even with different size members.

4 Operator

 • 

5.7K Posts

June 26th, 2013 04:00

Bhaskar1,

I assume you don't mean hypers, but symdevs or tdevs?

As for the answer: more small components in a META are always better for performance compared to a single large one.

Just make sure (using symdev show) that your devices in each component use different physical disks. As soon as you're seeing a spindel that you've seen before in that particular META, your META is too big. Make sure it's layed out over different disks.

4 Operator

 • 

5.7K Posts

June 26th, 2013 04:00

FYI: a hyper, slice or split is the piece on 1 physical disk (spindel). A few of these hypers/slices/splits together form a symdev or tdev.

Good read: http://www.50mu.net/2012/07/11/facilitate-the-conversation-say-what-you-mean-and-dont-make-assumptions/

2 Intern

 • 

244 Posts

June 26th, 2013 07:00

Hi Bhaskar,

We have moved your question to the Symmetrix Support Forum where it can be promptly addressed.

Thanks!

4 Operator

 • 

5.7K Posts

June 27th, 2013 01:00

Ah, indeed. In pools this works different. But for regular symdevs my story holds.

Sorry for the confusion. Quincy, you're right. As always.

1.3K Posts

June 27th, 2013 12:00

RRR, I WISH I was always right!

No Events found!

Top