Centralized Counters in a Distributed World Event Based Realtime Statistics using eTally
summary
Claims a real‑time sliding‑window counter system that handles ~10 million events/day in ~2.5 GB RAM, implemented as an eTally C++ core with an Erlang bridge. It keeps a FIFO queue plus per‑metric counters, supports leaderboards, means, variance and percentile buckets, and returns multi‑granularity (5m, 1h, 1d…2 months) aggregates in O(1) per update. Benchmarks show about 81k ops/s single-thread (208k with batching); one‑box production deployment works, disk‑backed queue and DB bootstrap provide durability options.
AI-written summary · about
chapters (4)
related talks
41mLessons Learned Implementing a Resilient, Distributed, Real-Time Architecture
15mLODIC: Logical Distributed Counting for Scalable File Access
17mCount-Based Abstractions for Performance Verification of Contention Points
19mDisco: Efficient Distributed Window Aggregation
26mDistributed Tracing: All the Warning Signs Were Out There!!!
01h 11mAsychronous design with Spring and RTI: 1M events per second
39mBuilding a Clustered Service in Go
34mChallenges in Measuring Erlang Application Performance