Display EBS actual disk information (gp3)
Currently, clusters created on AWS have gp3 volumes according to support.
However :
* the configuration displays max IOPS 6000 for M50 clusters even though documentation states that iops scale 3:1 with the volume size in GB (1 GB = 3 IOPS) regardless of disk type for M30+ clusters (true with gp2 anwyay, for gp3 that's an enforcement by Atlas)
Documentation link : https://docs.atlas.mongodb.com/customize-storage/#configuring-the-iops-rate
=> the configuration should at least display the actual IOPS (as it's also being used for alerting) - ideally it would allow us to have more IOPS than the minimum 3:1 enforcement.
- with gp2 we had no influence over the max throughput of the volume but we could know it was between 125 and 250 MiB/s. As per the AWS documentation , the max throughput is 250MiB/s depending on volume size ; a 334 GiB volume is the minimum to reach max throughput.
AWS doc : https://docs.aws.amazon.com/AWSEC2/latest/UserGuide/ebs-volume-types.html
With gp3, the minimum is 125MiB/s and the max 1000MiB/s and there's no information from Atlas regarding any rule they enforce for throughput.
According to Support, for our cluster, which has 3625 GB Storage size, the total throughput will be 587MiB/s. Which would be a Good News as we have twice as much throughput as before.
The documentation should explain the rule that is being enforced here.
Jean-Christophe
shared this idea
-
Martin
commented
Such a good point! I was wondering about the exact same detail of provisioned IOPS gp3 volumes...
Is Atlas using gp3 volumes at all or are they using io2 volumes when checking the "provisioned IOPS" checkbox?
The docs read as if it was io2, see: https://www.mongodb.com/docs/atlas/customize-storage
For io2 troughput would scale proportionally with IOPS, see: https://docs.aws.amazon.com/AWSEC2/latest/UserGuide/provisioned-iops.html