Kinetic
observability, reimagined
Home
Home Styles
Terminal
Blueprint
Signal
Product
Platform
Features
Integrations
Changelog
Content
Blog
Article Archive
Discover
All Categories
Our Writers
Pricing
Company
Company
About
Team
Contact
Pages
Account
Login
Register
Password Reset
Username Reminder
Content
Engineering
Single Article
Author
Discover
All Tags
Search
Landing
Sign in
Start free
Home
/
Blog
/ Engineering
CATEGORY
Engineering
Posts on performance, tracing, query internals and running Helix at scale.
ENGINEERING
The math behind our sampling-error bounds for approximate summarize
When HelixQL samples instead of scanning everything, how far off can the answer be? We publish the error bound, not...
Arjun Mehta · Jul 8
ENGINEERING
Profiling data as a fourth signal: storing stack traces as flame-graph deltas
Continuous profiling produces a full stack sample every 10ms per core. Storing each one independently would be absurd...
Lena Park · Jul 2
ENGINEERING
Backpressure: what happens when ingestion outruns compaction
A sustained ingest spike can outpace background compaction. Helix has to decide who slows down first, and it's a...
Priya Raman · Jun 30
ENGINEERING
How HelixQL resolves by clauses into group keys
A 'summarize... by' clause looks simple in HelixQL syntax, but resolving it against dynamic labels, computed...
Arjun Mehta · Jun 23
ENGINEERING
Cutting p95 by 40%: a checkout-api retro
How trace-driven debugging found a retry storm behind a healthy-looking dashboard.
Priya Raman · Jun 18
ENGINEERING
Compressing high-cardinality label dictionaries with a shared vocabulary
Two different metrics on the same pod share most of their label set. A per-metric dictionary wastes that overlap; a...
Priya Raman · Jun 16