Fullflash Performance when utilisation near 100% on DP+

[Fred-M]

Hello,
a customer buyed a new MSA2060 with 12x3,98TB SSD Drives and want to use a MSA-DP+ Pool for hosting his virtual machines.

He calculated two SSDs for stipe zones, but unfortunately forgot to consider the two additional SSDs for distributed spare capacity and so he ended up with 31,84TB useable space. (47,76TB - 4x3,98 = 31,81TB).

He needs to move 30TB on VM's to the new storage, would this be an massive performance impact?
I'am asking because from my experience the SSD performance slows down after reaching a certain amount of allocation (In this case 94% of 31,81TB), but i dont know how much impact this would have in an Full Flash Array with 12 SSDs with DP+.

Of course, another option would be to expand the pool with two additional 3,98TB SSDs. Therefore 39,8TB would be available and so only 75% utilisation. But would this make a difference from a performance perspective?

Perhaps someone had tested this scenario and got a answer,
Thanks in advance and best regards,
Fred

[ArunKKR · Accepted solution]

Hi,

The high threshold in this configuration should be around 99.4%.
This threshold is calculated based on available capacity of the pool minus 200 GB of reserved space.
Please execute the below command to identify the exact high threshold.

show pools

If the pool is overcommitted, the system will use write-through cache mode once virtual pool usage exceeds this high threshold.
This would still have a performance impact. The performance impact would be comparativley lower than a configuration with normal SAS drives.

By default, when a pool is created over commit would be enabled.

As long as the pool has more than 200GB free space there should not be any performance impact.

Ensure that the Pool usage does not exceed 99%.

You could create volumes of size not exceeding 99% of the Pool capacity.

[Fred-M]

Hello ArunKKR,

that was exactly what i needed to know, thank you very much for your detailed answer.

Best regards,
Fred