Add parallel streams throughput tests, and enable them in the EGW workflow #38027
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
|
/scale-egw |
944b9cf to
b5ebc1d
Compare
|
/scale-egw |
Let's not enable the egress gateway functionality at all in the baseline test, rather than enabling it but not configuring any policy. This ensures that baseline Cilium CPU/memory usage are not influenced by any Egress Gateway related tasks. Signed-off-by: Marco Iorio <marco.iorio@isovalent.com>
b5ebc1d to
e8e36e8
Compare
|
Rebased onto main to drop the commit already merged via #37981 |
|
/scale-egw |
christarazi
reviewed
Mar 7, 2025
marseel
approved these changes
Mar 7, 2025
Extend the CLI performance tests to support measuring the throughput with multiple parallel streams, rather than a single one. This allows to bypass possible single flow limits due to resource exhaustion or cloud provider limits, such as [1]. Unfortunately, netperf does not support multiple flows out of the box, hence we need to be a bit creative to run multiple instances and parse the resulting output. [1]: https://docs.aws.amazon.com/AWSEC2/latest/UserGuide/ec2-instance-network-bandwidth.html Signed-off-by: Marco Iorio <marco.iorio@isovalent.com>
e8e36e8 to
a4c8dfa
Compare
Following the extension of this logic to additionally support multiple parallel streams, let's add a unit test exercising the different paths, for increased confidence in its correctness, and to prevent future regressions. Signed-off-by: Marco Iorio <marco.iorio@isovalent.com>
So that we circumvent the single flow limitations imposed by AWS [1]. Additionally, let's change the node instance type to m5n.xlarge, as it supports up to 25Gbps burst bandwidth. [1]: https://docs.aws.amazon.com/AWSEC2/latest/UserGuide/ec2-instance-network-bandwidth.html Signed-off-by: Marco Iorio <marco.iorio@isovalent.com>
Gather the amount of traffic forwarded by the gateway node during the network performance test, and assert that it is close to 0 in the baseline test, and way higher in the EGW test. This serves as a sanity check that traffic is indeed flowing through the egress gateway when expected. Signed-off-by: Marco Iorio <marco.iorio@isovalent.com>
a4c8dfa to
f46aeb9
Compare
|
/test |
Single-stream flows are limited to 5 Gbps on AWS [1]. The multi-flow tests instead saturate the 25 Gb/s available bandwidth on these instances (except UDP + egress gateway, which tops at about 15 Gb/s due to CPU saturation). [1]: https://docs.aws.amazon.com/AWSEC2/latest/UserGuide/ec2-instance-network-bandwidth.html |
christarazi
approved these changes
Mar 7, 2025
viktor-kurchenko
approved these changes
Mar 10, 2025
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Extend the CLI performance tests to support measuring the throughput with multiple parallel streams, rather than a single one. This allows to bypass possible single flow limits due to resource exhaustion or cloud provider limits, such as [1]. Unfortunately, netperf does not support multiple flows out of the box, hence we need to be a bit creative to run multiple instances and parse the resulting output. Enable the new test as part of the egress gateway scale test, and change the node instance type to m5n.xlarge, as it supports up to 25Gbps burst bandwidth.
Please review commit by commit, and refer to the individual commit messages for additional details.
[1]: https://docs.aws.amazon.com/AWSEC2/latest/UserGuide/ec2-instance-network-bandwidth.html