So when BCV devices are synced from STD we are reaching the microcode limit sooner.
I think that the only limit you reach is the iops of the drives where either STD or BCV are. The code itself do have limits ... but I hope it's not your case
So what would be our approach to tune it so that BCV sync will be dont faster
You can ask to use SymMerge (or try to use Performance Manager) and look if your 32 drives where you have the BCV are "too busy" while you establish your devices. There are some different options if you want to "tune" your bcvs .. if you have some freespace, you can think about creating new BCV devices and use them instead of the old ones. Or you can use Optimizer and spread (by hand) the workload in a different way ..
It's however a very complex task and I don't think there is an automated tool that may be of help.
"The code itself do have limits ".. yes if write pending limit comes under code limit.
We were reaching the WP limit earlier too and a bin file modification (architecture modification) is done to spread the load of BCV devices .
Now there is an improvement ;WP limit is reaching sooner than earlier.(if it took 25 min earlier now it takes 15 min only)..but i am NOT clear if that is really helping here to improve the total time taken for the BCV sync&split since the WP limit is still applicable.
"The code itself do have limits ".. yes if write pending limit comes under code limit.
Usually when I talk about "Code limits" I talk about other limits .. WP is a function of cache .. The more cache you have, the higher WP limit goes .. It's not strictly bound to "the code" since changing the cache changes the WP limit.
If you have an open SR please send me the number since the forum isn't the right place where to talk about performances, neither in general nor in detail.
what's wrong with discussing performance issues ? I find it very informative even though a lot of time it pertains to specific customer configurations.
Forgive me .. But maybe I've been misunderstood .. It's not WRONG to talk about performances in general .. but may be misleading .. And it's not wrong to talk about specific issues ... but it's impossible to give always good advices without being able to "touch" the box itself, without being able to investigate the issue from "inside" the box. And the only way I know to have someone dial your box is to open a SR ..
xe2sdc
6 Operator
•
2831 Posts
273
0
Posted January 27th, 2008 03:00
reaching the microcode limit sooner.
I think that the only limit you reach is the iops of the drives where either STD or BCV are.
The code itself do have limits ... but I hope it's not your case
sync will be dont faster
You can ask to use SymMerge (or try to use Performance Manager) and look if your 32 drives where you have the BCV are "too busy" while you establish your devices.
There are some different options if you want to "tune" your bcvs .. if you have some freespace, you can think about creating new BCV devices and use them instead of the old ones. Or you can use Optimizer and spread (by hand) the workload in a different way ..
It's however a very complex task and I don't think there is an automated tool that may be of help.
Cheers!