queen dagger
If I use 7 E30 SSD disks (7x1TB, each disk can provide 500 IOPS and 60 MB/s) to form an aggregate on Azure CVO.
Can I use the following calculation to estimate how many IOPS and Throughput the aggregate can provide? if not, then what would be the correct calculation?
7 x 500IOPS = 3500 IOPS
7 x 60 MB/s = 420 MB/s
Thank you in advance for you inputs.
keen rover
Sounds right.
But keep in mind that aggr IOPS != frontend IOPS.
ONTAP isn't just JBOD...
queen dagger
But, I can only get 78MB/s, vs. 420MB/s (=7x60MB/s) if I use dd command on the client to test. I understand there would be some overhead on background, but still the difference looks too much. Any thing we can do to improve?
keen rover
Is there any latency in ONTAP?
queen dagger
In theory, is there a way we can calculate the IOPS and Throughput the aggregate can provide based on disks characteristics? That is what I would like to find out. Based on the calculation, the throughput should be 420MB/s but in reality we can only get 78MB/s, why?
vapid cypress
We don't provide maximum speeds for a system, it's a pretty complex question. For example, ONTAP taking "dd" will provide different results it's if coming from /dev/zero or /dev/random for example, as it does zero block detection and saves writes
keen rover
The disk IOPS will not equal total possible frontend IOPS. That's just a fact. There can be some sizing based upon your production workloads by the account team/cloud architects.