An investigation loop for your AI
October 27 2026, Newsletter no. 311 (this newsletter is available in html format at http://www.javaperformancetuning.com/news/news311.shtml )
The industry is accelerating issue response handling with AI. But you still need to guide the AI on the handling loop, AI on its own goes off on weird tangents or bizarre shortcuts that fake fixes. An SRE approach to performance issues is a good process for AI to follow:
1. *Is there a performance issue?* You should have monitoring and alerts, especially on the big items: latency, throughput, footprint.
2. *What is affected?* Observability is key to finding the components or request paths that are degrading
3. *Can we stabilize it?* Scale, roll back, reroute, failover.
4. *What caused it?* Correlate timeline to all events, find what changed, what are the triggers.
5. *How do we fix it?* Fix to remove the cause to prevent recurrence, benchmarking the change.
AI can be a powerful issue solver, but GIGO still applies so you need to make sure you keep it on track
Now on to all the usual newsletter list of links, tips, tools, news and talks, and as usual I've extracted all the tips into ( http://www.javaperformancetuning.com/news/newtips311.shtml ) this month's tips page.
A note from this newsletter's sponsor:
JProfiler
Get rid of your performance problems and memory leaks!
http://www.JavaPerformanceTuning.com/cgi-bin/urlon.pl?target=42
__News__
Java performance tuning related news
- Explaining how Elasticsearch accelerates vector search using the new Vector API and Foreign Function & Memory API at
https://www.youtube.com/watch?v=adFS-_sCTQ4
- Identity-free immutable value objects expected as a JDK 28 preview, allows denser layouts and can be handled and processed very efficiently by the JVM at
https://www.youtube.com/watch?v=lUB15M3mMLs
- Java performance improvements across JDK 25-28: G1 default everywhere, compact object headers, lazy constants, AOT improvements, etc at
https://www.youtube.com/watch?v=rNOzf12RORI
__Tools__
Java performance tuning related tools
- secure bridge to containerized JVMs for managing, storaging, retrieving, and analyzing JFR data, thread dumps, heap dumps, async-profiler sessions at
https://github.com/cryostatio/cryostat
- Identifying a memory leak by repeated class histograms, then a heap dump, then following a dominator tree from large retained objects back to Java code at
https://www.youtube.com/watch?v=z4jeFoT7SKY
- practical introduction to using HAT for GPU execution, progressing from thread-based up to higher-level tensor and tile APIs with profiling-driven optimizations at
https://www.youtube.com/watch?v=_1fi5GaWXxY
__Articles__
- Java Concurrency Deep Dive: Threads, ExecutorService, CompletableFuture (BiharGeeks/BiharGeeks) at
https://www.youtube.com/watch?v=H602ia22jmo
- High Performance Java Native AI with CUDA and the Vector API (Zoran Sevarac/Devoxx) at
https://www.youtube.com/watch?v=8NURE3I4AJQ
- Java Secrets for Elegant Low Latency (Peter Lawrey/Today Software Magazine) at
https://www.youtube.com/watch?v=xtz5bxyCpog
__The latest performance tips__
All the performance tips extracted from the listed articles at http://www.JavaPerformanceTuning.com/news/newtips311.shtml
This email has been sent to you because you have subscribed to the Java Performance Tuning newsletter. If you prefer not to receive any future newsletters, please use the URL http://www.javaperformancetuning.com/cgi-bin/subscribe.pl?unsubscribe=[subscriber]
Jack Shirazi http://www.javaperformancetuning.com/contactUs.shtml Java(TM) is a trademark or registered trademark of Oracle Corporation in the United States and other countries. JavaPerformanceTuning.com is not connected to Oracle Corporation and is not sponsored by Oracle Corporation.