Recommended disk repository for TDP for SAP HANA

aayala85

ADSM.ORG Member
Joined
Nov 4, 2018
Messages
14
Reaction score
0
Points
0
PREDATAR Control23

Hello,

What is the recommended disk storage pool repository for TDP for SAP HANA?

We have 40 TB disk with easy tier and 80 TB HDD disk.

container pool, sequential disk pool, or disk ramdom acces?

Best regards,

AjM
 
PREDATAR Control23

Easy tier and Spectrum Protect don't mix.

Whatever you do, don't put the Spectrum Protect Server database on easy tier, and don't use container pools if you plan to put them on easy tier.

That leaves you with random pool, but not sure it's a good match for easy tier either.

The issue is that easy tier works well for an application that keeps adding new data, but rarely accesses old data. Spectrum Protect works in a random matter. New data is ingested daily, and old data is expired daily. So need to access both old and new on a constant basis. Easy tier just works against you with such an environment.
 
PREDATAR Control23

Easy tier and Spectrum Protect don't mix.

Whatever you do, don't put the Spectrum Protect Server database on easy tier, and don't use container pools if you plan to put them on easy tier.

That leaves you with random pool, but not sure it's a good match for easy tier either.

The issue is that easy tier works well for an application that keeps adding new data, but rarely accesses old data. Spectrum Protect works in a random matter. New data is ingested daily, and old data is expired daily. So need to access both old and new on a constant basis. Easy tier just works against you with such an environment.

OK, thanks marclant

So between random disk devclass and sequential file devclass, wich will be better for TDP for SAP HANA?

Best regards,

AjM
 
PREDATAR Control23

OK, thanks marclant

So between random disk devclass and sequential file devclass, wich will be better for TDP for SAP HANA?

Best regards,

AjM
With easy tier, hard to tell. If you do sequential, that "may" work better. Old volumes will be tiered to slow disks, and hopefully new volumes will be created on fast disks for your backups to write to.

My recommendations is test both ways, both backups and restores for at least a few days.
 
PREDATAR Control23

With easy tier, hard to tell. If you do sequential, that "may" work better. Old volumes will be tiered to slow disks, and hopefully new volumes will be created on fast disks for your backups to write to.

My recommendations is test both ways, both backups and restores for at least a few days.

OK, thanks

And if the easy tier disk is put aside and only HDD disk is used,
which one do you think would be the best option for the best performance from random access or sequential file?

Best regards,

AjM
 
PREDATAR Control23

There shouldn't be much difference either way. Random has less management. In either cases, following best practices will be what's going to affect the performance the most: https://www.ibm.com/support/knowledgecenter/SSEQVQ_8.1.8/perf/r_srv_hw_disks.html

If you are not using easy tier, then you could potentially use container pools, but if you do, make sure to follow the Blueprint: http://ibm.biz/IBMSpectrumProtectBlueprints
OK, thanks marclant.

If we go with random disk, what volume size do you suggest?

Best regards,

AjM
 
PREDATAR Control23

It's in the best practices link I provided earlier, check the explanation to this question:
For storage pools that use DISK device classes, have you determined how many storage pool volumes to put on each file system?

And to this:
Did you create your storage pools to distribute I/O across multiple file systems?
 
Top