Mirrorclone, performance impact

hpuserabc — Tue, 07 Oct 2008 03:34:34 GMT

is there any performance impact when using mirrorclone ?

could perfmon show the significant difference ?

i would like to know the difference but it should based on data/numbers not merely theory.

any advice on how to do the testing will be appreciated

thx in adv

Re: Mirrorclone, performance impact

Uwe Zessin — Tue, 07 Oct 2008 04:10:06 GMT

Sure there will be a performance impact - such actions are never free. Whether it is noticable for the end-user or can be absorbed by caching depends on many things (as usual ;-)

We cannot give you an exact number (mirror cloning will cost you N% of X) as every array is different. EVAperf can give you information about latencies down to the individual virtual disk. No offence meant, but I think you need to get the numbers from your own installation.

Set up a virtual disk, wait until the initial zeroing has completed and run some tests with "iometer". Set up a mirror-clone, wait until the initial copy has completed and run the same tests. Fracture the mirror-clone (stop the continuous sync, but keep a differential bitmap) and run some more tests (should include writes).

Re: Mirrorclone, performance impact

hpuserabc — Tue, 07 Oct 2008 08:19:40 GMT

thx uwe

topic Re: Mirrorclone, performance impact in HPE EVA Storage

Mirrorclone, performance impact

Re: Mirrorclone, performance impact

Re: Mirrorclone, performance impact