Performance regression in aws-sigv4 1.5.0

[johnduhart]

Describe the bug

While doing performance profiling of our application which manually performs AWS SigV4 signing (for Amazon Neptune), we noticed a measurable amount of time being spent in Signer.use_crt?.

Looking closer, it appears to be calling require on each invocation.

We deployed a build with aws-sigv4 rolled back to 1.4.0, and we noticed a large improvement on the lower latency percentiles for calls to Amazon Neptune (latency includes signing, the HTTP request, and serialization).

aws-sigv4 version	1.4.0	1.5.0
P25	5ms	16ms
P50	5ms	17ms
P75	6ms	19ms
P90	7ms	22ms

Memoizing the use_crt? call would fix this issue. Installing and using the AWS CRT might also improve things, but given it's labeled as a developer preview, we decided to just rollback for now.

Expected Behavior

The same performance characteristics as 1.4.0

Current Behavior

Worsened performance.

Reproduction Steps

Benchmark the performance of sign_request between 1.4.0 and 1.5.0.

Possible Solution

No response

Additional Information/Context

No response

Gem name ('aws-sdk', 'aws-sdk-resources' or service gems like 'aws-sdk-s3') and its version

aws-sigv4

Environment details (Version of Ruby, OS environment)

Ruby 2.7.6, Linux

[alextwoods]

Thanks for the detailed report - I'll work on a fix!

[github-actions]

⚠️COMMENT VISIBILITY WARNING⚠️

Comments on closed issues are hard for our team to see.
If you need more assistance, please either tag a team member or open a new issue that references this one.
If you wish to keep having a conversation with other community members under this issue feel free to do so.