EVA6400 along with BC snapshot

[zxcv]

Hi ,

WE have around 40 clients for which 40 applcn and 40 databases.

Out of which 16 clients have database of around 400gb ( 4 Vdisk of 100gb ) RAID 1 each for which we take Snapshot HP RSM.

We present the same to two different backup servers each 8 clients.

Now the issue is when we run 6 BC in a gap of 5 min our whole server accesing this EVA gets hanged, i mean bdf hangs our application queues hang up and afetr 1 hr when the levelling is complete we are back to normal.

what needs to be done to avoid this hangup ?

[Johan Guldmyr]

Hi.

One thing you could do is to run one BC at the time.

Then two, then three, until you find the limit.

Do you take a normal snapshot?

Have you looked into snapclone or mirrorclone?

Maybe those would work better in your scenario.

[zxcv]

Hi Johan ,

We take snapclone and not snapshot..am sorry for typo mistake.

Regarding mirrorclone i dont think it fits our requirement.

[Sheldon Smith]

---

@zxcv wrote:

Hi Johan ,

We take snapclone and not snapshot..am sorry for typo mistake.

Regarding mirrorclone i dont think it fits our requirement.

---

No doubt you are right. Of course, if we knew your requirements as well, perhaps we could better advise you. What is the underlying problem/scenario you want to solve?

And since you are already doing snapclones, mirrorclones may fit your requirement.

[zxcv]

Hi Sheldon ,

My requirement is that i must be able to take BC simultaneously and my I/O should not get hampered .

As far as exact requiremnt is concerned we have 16 large databases for which export is out of scope.

So we take a snapclone of existing one and then mount it on a backup server whereon we take an export of the same. ( We put the database in begin and end backup mode )

Do we have a software for this kind of requirement ?

[Sheldon Smith]

So instead of snapclones, use mirrorclones.

For a database,

1. begin backup mode,
2. fracture the clone(s),
3. end backup mode. You now have copies ready to go.
4. Present and mount clone(s) to the backup server.
5. Backup.
6. Some time after the backup is complete, unmount, unpresent and start the resync(s).

[zxcv]

Hi Sheldon ,

Let me explain you my exact scenario.

We have 16 big clients whose database export is out of scope.

so we take BC for bank1 at 9.00 pm everyday which we term it as pre-eod bkp , we present it to a bkup server. On bkup server we take export backup of that client.

then say at 3 .00 am we take another BC of the same client which overwrites the existing copy from bkp server ( Note : We only take this BC when we are sure that all our imp jobs of pre-eod are completed without any issues. ) and then we take post-eod bkp for that client.

Here we are saving our backup time during pre as well as post , also we have a post copy available on the bkp server always the next day.

Until now everything was fine , but recently the interval at which we take BC has reduced so much that we have to fire BC for all 16 banks at only 5 min gap , which chokes my I/O , bdf gets hanged all imp processes go in loop ( take more time to finish )

so i am looking for an alternate solution for snapclone ,

can mirroclone as suggested by you meet this requirement ?

[Sheldon Smith]

Mirrorclone may help, but I think you are to the point you should be talking with your HP Storage Advisor, or at least HP pre-sales consultants to help you come up with a solution that HP will support.

[zxcv]

Hi Sheldon ,

I will surely log  a call with HP for the same.

Thank you very much for your support.