aws-ec2: add m6i as preferred instance type#5327
aws-ec2: add m6i as preferred instance type#5327openshift-merge-robot merged 1 commit intoopenshift:masterfrom
Conversation
|
Skipping CI for Draft Pull Request. |
openshift-ci
Bot
added
the
do-not-merge/work-in-progress
|
The cluster install with no issues[1], and MAPI has no restrictions to replace nodes for the new type[2] [1] provisioned by installer: [2] Machine with new size (replaced in a existing cluster): |
staebler
left a comment
staebler
left a comment
There was a problem hiding this comment.
This seems reasonable to me. Let me discuss with my team to make sure that I am not missing some wrinkle on why we would not want to do this.
AWS recently launched the m6i instances that is a newer generation of m5 (default AWS IPI). The m6i offer up to 15% improvement in price/performance versus comparable fifth-generation instances. The new instances are powered by the latest generation Intel Xeon Scalable processors (code-named Ice Lake) with an all-core turbo frequency of 3.5 GHz.
Compared to M5 instances using an Intel processor, this new instance type provides:
- Up to 15% improvement in compute price/performance.
- Up to 20% higher memory bandwidth.
- Up to 40 Gbps for Amazon Elastic Block Store (EBS) and 50 Gbps for networking.
- Always-on memory encryption.
The m6i instance is available in 8 of 21 regions, so clusters running on those regions can take advantage of that new instance type.
count(m5.xlarge) count(m6i.xlarge) count(m5_diff_m6i)
region
af-south-1 3 0 3
ap-east-1 3 0 3
ap-northeast-1 3 2 1
ap-northeast-2 4 0 4
ap-northeast-3 3 0 3
ap-south-1 3 0 3
ap-southeast-1 3 3 0
ap-southeast-2 3 0 3
ca-central-1 3 0 3
eu-central-1 3 2 1
eu-north-1 3 0 3
eu-south-1 3 0 3
eu-west-1 3 3 0
eu-west-2 3 0 3
eu-west-3 3 0 3
me-south-1 3 0 3
sa-east-1 3 0 3
us-east-1 5 5 0
us-east-2 3 3 0
us-west-1 2 2 0
us-west-2 6 4 2
The point of attention is the regions that are not deployed that type across all zones: ap-northeast-1, eu-central-1 and us-west-2. But afaik it should not be a problem as the algorithm to choose the better instance should get the instance more available in the region.
Changing the default AWS IPI ec2 to m6i.xlarge and the EBS volume type to gp3 on control planes could be a "easy win" for performance and costs.
mtulio
force-pushed
the
aws-ec2-m6i-preferred
branch
from
05d2549 to
b918810
Compare
staebler
left a comment
staebler
left a comment
There was a problem hiding this comment.
Please remove the draft status when you are comfortable.
/lgtm
|
Note that CI is hard-coded to use m5 instances. |
|
[APPROVALNOTIFIER] This PR is APPROVED This pull-request has been approved by: staebler The full list of commands accepted by this bot can be found here. The pull request process is described here DetailsNeeds approval from an approver in each of these files:
Approvers can indicate their approval by writing |
openshift-ci
Bot
added
the
approved
|
Cool, thanks to review @staebler . I am also converting to a PR. I can't see any issue to backport it to current releases (4.6+), what do you think? |
openshift-ci
Bot
removed
the
do-not-merge/work-in-progress
No, I don't think this meets the bar for backporting. Using 5th-gen instances does not prevent a user from installing and using OpenShift. And the user has the option to set the instance type themselves if they want to take advantage of the 6th-gen instances. |
|
/retest-required Please review the full test history for this PR and help us cut down flakes. |
4 similar comments
|
/retest-required Please review the full test history for this PR and help us cut down flakes. |
|
/retest-required Please review the full test history for this PR and help us cut down flakes. |
|
/retest-required Please review the full test history for this PR and help us cut down flakes. |
|
/retest-required Please review the full test history for this PR and help us cut down flakes. |
|
@mtulio: The following tests failed, say
Full PR test history. Your PR dashboard. DetailsInstructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the kubernetes/test-infra repository. I understand the commands that are listed here. |
|
/retest-required Please review the full test history for this PR and help us cut down flakes. |
3 similar comments
|
/retest-required Please review the full test history for this PR and help us cut down flakes. |
|
/retest-required Please review the full test history for this PR and help us cut down flakes. |
|
/retest-required Please review the full test history for this PR and help us cut down flakes. |
|
/retest-required Please review the full test history for this PR and help us cut down flakes. |
3 similar comments
|
/retest-required Please review the full test history for this PR and help us cut down flakes. |
|
/retest-required Please review the full test history for this PR and help us cut down flakes. |
|
/retest-required Please review the full test history for this PR and help us cut down flakes. |
4 participants
AWS recently launched the m6i instance that is a newer generation of m5 (default AWS IPI). The m6i offers up to 15% improvement in price/performance versus comparable fifth-generation instances. The new instances are powered by the latest generation Intel Xeon Scalable processors (code-named Ice Lake) with an all-core turbo frequency of 3.5 GHz.
Compared to M5 instances using an Intel processor, this new instance type provides:
The m6i instance is available in 8 of 21 regions, so clusters running on those regions can take advantage of that new instance type.
The point of attention is the regions that are not deployed that type across all zones: ap-northeast-1, eu-central-1, and us-west-2. But afaik it should not be a problem as the algorithm to choose the better instance should get the instance more available in the region.
Changing the default AWS IPI ec2 to m6i.xlarge and the EBS volume type to gp3 on control planes could be an "easy win" for performance and costs.