Bug 2044340

Summary: AWS | gp3 volume provide only default performance
Product: [Red Hat Storage] Red Hat OpenShift Data Foundation Reporter: Avi Liani <alayani>
Component: ocs-operatorAssignee: Mudit Agarwal <muagarwa>
Status: CLOSED NOTABUG QA Contact: Elad <ebenahar>
Severity: unspecified Docs Contact:
Priority: unspecified    
Version: 4.10CC: etamir, mmuench, muagarwa, nigoyal, odf-bz-bot, owasserm, sostapov
Target Milestone: ---Keywords: Performance
Target Release: ---   
Hardware: Unspecified   
OS: Unspecified   
Whiteboard:
Fixed In Version: Doc Type: If docs needed, set a value
Doc Text:
Story Points: ---
Clone Of: Environment:
Last Closed: 2023-08-08 08:35:45 UTC Type: Bug
Regression: --- Mount Type: ---
Documentation: --- CRM:
Verified Versions: Category: ---
oVirt Team: --- RHEL 7.3 requirements from Atomic Host:
Cloudforms Team: --- Target Upstream Version:
Embargoed:

Description Avi Liani 2022-01-24 12:16:26 UTC
Description of problem (please be detailed as possible and provide log
snippests):
while deploying ODF 4.10 on AWS with default gp3 backend volumes, we are getting only 3K IOPS and 125 MB/Sec throughput - this is the defaults from AWS.
 

Version of all relevant components (if applicable):

OCP : 4.10
OCS : 4.10

Does this issue impact your ability to continue to work with the product
(please explain in detail what is the user impact)?
default performance is 50% of default gp2 (2 TiB volumes)


Is there any workaround available to the best of your knowledge?
yes - by deploying manually user can setup the gp3 to work with better performance

Rate from 1 - 5 the complexity of the scenario you performed that caused this
bug (1 - very simple, 5 - very complex)?
1

Can this issue reproducible?
yes

Can this issue reproduce from the UI?
yes

If this is a regression, please provide more details to justify this:


Steps to Reproduce:
1. deploy OCP 4.10 on AWS
2. deploy ODF 4.10 on default gp3 storageclass
3. run FIO and mesure performance


Actual results:

the performance is 50% less than performance on gp2 storageclass

Expected results:

better performance from gp2

Additional info:

for deploying with "high-performance" gp3 devices, we need  to create new storage class like :

kind: StorageClass
apiVersion: storage.k8s.io/v1
metadata:
  name: gp3-iops-10k
  annotations:
    storageclass.kubernetes.io/is-default-class: "true"
allowVolumeExpansion: true
provisioner: ebs.csi.aws.com
volumeBindingMode: WaitForFirstConsumer
parameters:
  type: gp3
  iops: "16000"
  throughput: "500"

or setting to our storageclass the parameters :

parameters:
  type: gp3
  iops: "16000"
  throughput: "500"

Comment 2 Nitin Goyal 2022-01-24 14:25:19 UTC
Moving it to the ocs-operator

Comment 3 Yaniv Kaul 2022-01-25 15:58:18 UTC
The reason it's default is that this is what you get from free. I think we can expect ODF to change this when we fully understand the benefits of gp3 (which I'm sure we have), with or WITHOUT changing the CPU+memory of the OSDs!

Comment 4 Jose A. Rivera 2022-02-01 16:24:32 UTC
Since we're still exploring the utility/viability of gp3 for our backend volumes (as well as the pricing differences), moving this out to ODF 4.11. We will still continue our testing, of course.