feat(v2): parallel batch processing

[jeromevdl]

Issue #, if available: #1540

Description of changes:

batch module can now process items in parallel rather than in sequence. Introduction of a new method processBatchInParallel in BatchMessageHandler. Works with SQS, Kinesis, DDB Streams but not SQS FIFO (does not make sense).

Checklist

• Meet tenets criteria
• Update tests
• Update docs
• PR title follows conventional commit semantics

Breaking change checklist

RFC issue #:

• Migration process documented
• Implement warnings (if it can live side by side)

By submitting this pull request, I confirm that you can use, modify, copy, and redistribute this contribution, under the terms of your choice.

[github-actions]

💾 Artifacts Size Report

Module	Version	Size (KB)
powertools-common	2.0.0-SNAPSHOT	9.59
powertools-serialization	2.0.0-SNAPSHOT	17.23
powertools-logging	2.0.0-SNAPSHOT	33.07
powertools-logging-log4j	2.0.0-SNAPSHOT	20.69
powertools-logging-logback	2.0.0-SNAPSHOT	16.91
powertools-tracing	2.0.0-SNAPSHOT	14.01
powertools-metrics	2.0.0-SNAPSHOT	14.05
powertools-parameters	2.0.0-SNAPSHOT	17.49
powertools-validation	2.0.0-SNAPSHOT	19.94
powertools-cloudformation	2.0.0-SNAPSHOT	16.58
powertools-idempotency-core	2.0.0-SNAPSHOT	34.63
powertools-idempotency-dynamodb	2.0.0-SNAPSHOT	12.37
powertools-large-messages	2.0.0-SNAPSHOT	17.45
powertools-batch	2.0.0-SNAPSHOT	21.49
powertools-parameters-ssm	2.0.0-SNAPSHOT	10.70
powertools-parameters-secrets	2.0.0-SNAPSHOT	9.90
powertools-parameters-dynamodb	2.0.0-SNAPSHOT	11.95
powertools-parameters-appconfig	2.0.0-SNAPSHOT	11.45

[scottgerring]

Nice! Some feedback inline

[sonarcloud]

Quality Gate passed

Issues
6 New issues
0 Accepted issues

Measures
0 Security Hotspots
No data about Coverage
0.0% Duplication on New Code

See analysis details on SonarCloud

[humanzz]

Hi folks,

I'm quite happy to see you're working on this. I know my team would love such ability.

I have not read the PR yet, but thought I'd leave a couple of comments to consider - if you haven't already.

In some of our Lambda functions, where we use powertools (more specifically logging, tracing), we do leverage concurrency to speedup our lambda requests. Depending on the use case, this concurrency is achieved using (1) AWS SDK async clients (which return Completable futures) (2) Other types of clients that might have their thread pools (3) Explicit executor services (and sometimes virtual threads from Java 21).

The core challenges we tend to see are in logging/tracing

1. In the beginning of handling a request, we like to add some properties to both powertools logging context and metrics context. This allows us to easily debug issues, and trace requests, via business-relevant identifiers
2. We see challenges with xray tracing e.g. either no trace emitted, or trace having wrong parent. All of this is likely due to xray library depending on thread locals to have this lineage tracking, which becomes very challenging with existing thread pools we don't control. Most of the time we end up having to write some custom pieces of code to try to pass relevant info across threads when possible
3. We also see challenges with logging context, again with thread pools we don't control

I kinda feel, this change, is likely the first time that powertools is handling concurrency. I wonder, if as part of that, or as follow ups, challenges with concurrency when it comes to tracing/logging can maybe be looked at, to see if there's any level of help powertools can provide.

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 76.08%. Comparing base (82d4b30) to head (789b91b).
Report is 75 commits behind head on v2.

❗ Current head 789b91b differs from pull request most recent head 395f37a

Please upload reports for the commit 395f37a to get more accurate results.

Additional details and impacted files

@@              Coverage Diff              @@
##                 v2    #1620       +/-   ##
=============================================
- Coverage     89.79%   76.08%   -13.71%     
- Complexity      406      426       +20     
=============================================
  Files            44       40        -4     
  Lines          1274     1560      +286     
  Branches        165      240       +75     
=============================================
+ Hits           1144     1187       +43     
- Misses           88      288      +200     
- Partials         42       85       +43     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[scottgerring]

We see challenges with xray tracing e.g. either no trace emitted, or trace having wrong parent. All of this is likely due to xray library depending on thread locals to have this lineage tracking, which becomes very challenging with existing thread pools we don't control. Most of the time we end up having to write some custom pieces of code to try to pass relevant info across threads when possible

Hey @humanzz ! Sorry for the slow reply; it's been busy over here.
@jeromevdl and I were speaking about this this morning. The error mode you describe from your experience with pooling - missing traces and strange parents - is the sort of thing I am a keen to see we avoid here.

I think the broader issue of "things that need to propagate information across thread boundaries don't work well" is something we may well be positioned to help with - certainly with logging and tracing at least. I have opened #1670 to track this.

[scottgerring]

@jeromevdl how about we merge this without the fix for x-ray and handle that on another PR?

Given the x-ray and log correlation is likely going to be a bigger thing, and we are talking about a SNAPSHOT here, I think the downside of merging this into v2 is low - some x-ray traces won't correlate well on a preview release - and it will let us deal with the bigger issue asynchronously rather than scope-creeping this PR.

[jeromevdl]

@jeromevdl how about we merge this without the fix for x-ray and handle that on another PR?

Given the x-ray and log correlation is likely going to be a bigger thing, and we are talking about a SNAPSHOT here, I think the downside of merging this into v2 is low - some x-ray traces won't correlate well on a preview release - and it will let us deal with the bigger issue asynchronously rather than scope-creeping this PR.

I agree, let's review like it is today and create a new issue for the multithread support

[jeromevdl]

See #1671

[scottgerring]

@jeromevdl i'll review this tomorrow morning

[sonarcloud]

Quality Gate passed

Issues
11 New issues
0 Accepted issues

Measures
0 Security Hotspots
0.0% Coverage on New Code
0.0% Duplication on New Code

See analysis details on SonarCloud

[scottgerring]

🤝