MSA 2050 - different sized power of 2 RAID VDGs vs balanced VDGs for 2 pools

[Michael Chan (TDM)]

I have an MSA 2050 with 14 x 15K 900GB HDDs.  From the MSA Gen5 Virtual Storage Technical Reference Guide, I follow the power of 2 for number of drives in a RAID 5 / 6 VDG and create an 8+1 RAID 5 VDG for pool A and a 4+1 RAID 5 VDG for pool B.  Alternatively, I can also create two 6+1 RAID 5 VDGs and assign one to pool A and the other to pool B to balance the load across the two controllers.

Since the drives are HDDs that do not come close to exercising the bandwidth limits of the controllers, I want to use the first option of two different power of 2 size VDGs for the two pools.  What are considerations that might justify using the second option of two same but not power of 2 size VDGs for the two pools instead of the first option?

[Shawn_K]

Hello,

Your question is hard to answer since you have just given how many drives you have but not what your storage will be used for. Databases, VDI application, test and dev deployment, scratch data use, etc? You have not provided enough detail as to whether you need to configure for access, swapping, large data transfers, or long term storage whihc makes the question hard to answer.

The one item that concerns me in both of your scenarios is the use of RAID5. Both configurations you have placed all available drives into RAID5 disk-groups without consideration of spares. While you design and consider your system you should also take into account that you will eventially encounter a drive failure - so build in redundancy now at the start.

I personally prefer the use of RAID6 but if your needs fit better into RAID5 then consider what your usage and application needs are and design up from there being sure to consider at least 2 spares in the configuraiton. From there I would consider what the storage is going to be used for and design to balance workload across the system.

Good luck!

Cheers,
Shawn

I work for Hewlett Packard Enterprise. The comments in this post are my own and do not represent an official reply from HPE. No warranty or guarantees of any kind are expressed in my reply.

[Michael Chan (TDM)]

Hello Shawn,

Thanks for recommendation of adding spare drive.

Workload is transaction processing development system. RAID 5 was selected as compromise between better performance and lower storage efficiency of RAID 1, and worse write performance of RAID 6.

Understand the power of 2 guideline for RAID 5 / 6 disk groups helps sequential access workload more than random access workload.  Understand different sized disk group will have different performance so advantage is for more similar performance between two pools of storage is disk group size is same for both pools.  Thoughts on whether one might outweigh the other for workload with random accesses?

Regards,

Mike

[Mulesh_Vedant]

Hi Shawn,

What is the impact of block size configured by the application v/s number of drives in a RAID group. 

For instance, if every write is 8K block size, should I consider 8d+2p configuration? 

Thanks

Mulesh