Art’s Substack

Art’s Substack

Home
Notes
Archive
About
43 Bytes per Series: How I Compressed OTLP labels with Packed KeySets
From native strings to packed keysets, this data-driven OTLP LabelSetStore shows how to cut memory to 43 bytes/series at scale
 • Art
Stop Allocating Per Label: A Data‑Driven Rust SymbolTable for OTLP/TSDB
I measured it instead of arguing: Arc<str> vs arena vs SmolStr vs GermanStr on OTLP-like data. Allocation counts and fragmentation tell the real story.
 • Art
The GIL Was Your Lock
Free-threaded Python is the biggest change to the ecosystem in a decade. While it unlocks massive performance potential, it also removes the "accidental…
 • Art
Free-Threading Python vs Multiprocessing: Overhead, Memory, and the Shared-Set Meltdown
Free-threading makes CPU threads real - but should you ditch multiprocessing? Benchmarks across Linux/Windows/macOS exposes spawn tax, RAM tax, and a…
 • Art
Python Threads: GIL vs Free-Threading
For years, Python threads don’t speed up CPU-bound work has been one of those truths that felt less like a technical statement and more like a life…
 • Art
When Profiler Deadlocks. Debugging a Rust Service Stuck in futex with GDB
Production Rust service deadlocked. 😱 We dove deep with gdb into a futex wait, traced it to jemalloc, and found a nasty bug in profiler's signal…
 • Art
Building a Custom Android System with eBPF Support
Customize Linux Kernel for Android to run eBPF
 • Art
Art’s Substack
Art’s Substack
My personal Substack
Recommendations
The AI Positive Substack
The AI Positive Substack
Paul Datta

Art’s Substack

AboutArchiveRecommendationsSitemap
© 2026 Art · Privacy ∙ Terms ∙ Collection notice
Start your SubstackGet the app
Substack is the home for great culture