This is by far the best technology book I’ve read in a long time. Get yourself a copy amazon.co.uk/Learning-eBPF-…
Richard Startin
1,193 posts
- Debugging latency with profilers
- I missed this talk on Java at AWS at JFokus last week, but just watched the recording and highly recommend watching it. youtu.be/7DKQ8OdGfmk?si…
- We’ve been working on contextualising profiling data at Datadog. Here’s an example of a ZGC allocation stall in what should be a CPU bound workload. The view, arranged as a timeline per thread, is scoped to the request this allocation stall delayed.
- Object Introspection from Facebook uses dwarf debug data to recover object graphs from live processes, which appears to offer unprecedented insight into memory occupancy
- Replying to @yagiznizipli and @fleming_mattAnd tell them to please use continuous profiling so they can know where their bottlenecks are, so much more satisfying than pontification and supposition.
- How crash handling works in async-profiler
- I’m leaving Datadog at the end of next month and plan to spend some time on personal projects, I’ll probably look at finding another job next September.
- SWAR indexOf implemented in elasticsearch. It makes me wonder if there ought to be a utility method in Arrays (which could have a much better intrinsic implementation) rather than have every Java project implement it.
- I’ll be talking about balancing software performance and complexity, techniques and tools to find and solve performance problems, and how hashing strings got a lot faster in Java 21 at QCon London on April 8th.
- Replying to @mariofuscoYou copied that null check without understanding why it does what it does, and as a result your code IS GARBAGE. AGAIN.
- Replying to @joseph_h_garvinI found protobuf penalises writers when there is nesting in the schema, which can be too expensive when you want tooling to be more or less invisible, not sure why it’s so revered as a format.
- I expect LLMs to eventually increase the status of the maintenance programmer. Partial or complete automation of code creation will lead to regulatory pressure for *someone* to understand systems in certain critical domains- a skill set not unlike maintenance programming.
- I fixed a segfault in AsyncGetCallTrace (my first OpenJDK commit) before I went on holiday and by the time I got back the Amazon Corretto team had backported it 😍
