As of today, @OptimyzeCloud is a part of @elastic :-) -- we're looking forward to integrating our tech and bringing frictionless continuous whole-system profiling to the entire Elastic ecosystem :-) - and combining it with logs, metrics, traces, and alerting!
Allright. As of today, we (as in @OptimyzeCloud) are all part of @elastic. What a crazy ride. I will take a couple days off, and then there's a mountain of work to tackle :-) -- from ARM64 support to .Net to making sure you can continuously profile on-prem and integrate ...
New blog post by @seanhn! prodfiler.com/blog/optimisin… Performance analysis of K2, an optimising compiler for eBPF, with Prodfiler.
Outcome: 1.4x-1.9x speedup via a new memory allocator, mods to allow autovectorisation, and PGO/LTO.
This talk from @rakyll provides a great overview of using eBPF for observability in microservices which I've never had to think about before but now spend a lot of time thinking about thanks to Astra DB
youtu.be/-I29Cuj2nbI
.@OptimyzeCloud CEO Thomas Duillen wrote this fantastic blog about delving into how kubelet was eating up CPU and IOPs. Read this, then register for #P99CONF, the free online conference where he'll be speaking. p99conf.io/2021/09/22/the-my…
"what is clear is that this tool is super awesome and provides an unseen level of visibility" <- anonymous user feedback of the best kind :-)
prodfiler.com :-)
If you like kernel deep dives, you'll like this. @vic_mic_ did a followup for my demand paging piece, with a performance focus, covering a LOT that I missed.
My favorite part: the clever page-table fu that triggers pathological kernel allocation behavior via a top-down scan. 🥋
New blog post: @vic_mic_ takes Prodfiler down into the depths of the Linux memory management system ... to examine "The OOMsday device".
prodfiler.com/blog/the-oomsd…
New blog post: @vic_mic_ takes Prodfiler down into the depths of the Linux memory management system ... to examine "The OOMsday device".
prodfiler.com/blog/the-oomsd…
New blog post: @vic_mic_ takes Prodfiler down into the depths of the Linux memory management system ... to examine "The OOMsday device".
prodfiler.com/blog/the-oomsd…
fun things prodfiler by @OptimyzeCloud tells me about my servers. It takes more CPU cycles to handle logging mechanism then to serve my cluster of websites :P. Also recently encountered a scenario where server was functional till ELK filebeat was started then the beat was gone
FYI: we wrote a further blog post on the matter with more analysis and a neat example on fuzzer performance issues: "A stroll down fuzzer optimisation lane and why instrumentation policies matter" blog.envoyproxy.io/a-stroll-…
How much carbon do you expend on fuzzing?
Quickly tried @OptimyzeCloud Prodfiler against an HTTP3 google oss-fuzz h2o target. The graphs are responsive and I like the carbon/cost reporting. Looks pretty cool.
Still syncing. Up to block about 605000 the flamegraphs report that about half the time was spent in background LevelDB compaction tasks (?). The other half was spent in block validation.
We are very excited that Prodfiler was just announced to the world! 👏👏
Our team is thrilled to be part of this project and collaborate with the @OptimyzeCloud team in such an innovative solution. Please look forward to it! 😎
#dataviz#devops
It is alive! Introducing Prodfiler, the world's first frictionless whole-fleet whole-system continuous profiler: prodfiler.com/blog/introduci… -- profile all your code, everywhere, all the time.
Try it today :-)