A new single-binary event log called Ayder runs in C and uses plain HTTP. You can publish and read messages with curl, sidestepping heavy clients or the JVM. A two-minute demo shows it surviving a SIGKILL crash and resuming without losing data. The creator is seeking early design partners for real-world ingestion workloads.

Under the hood, Ayder uses Raft consensus for sync-majority writes and append-only files for durability. Benchmarks on a three-node cluster show sustained throughput near 50K msg/s and client P99 latency around 3.46ms. After an unclean kill, recovery reportedly takes 40–50 seconds with millions of offsets, avoiding multi-hour restarts common in heavier stacks.

Compared with Kafka and Redis Streams, Ayder bets on HTTP to cut operational complexity. No ZooKeeper, no JVM tuning, and no async-only replication trade-offs. It supports consumer groups, KV operations with CAS and TTL, and stream processing across formats. The pitch is familiar: deliver Kafka-grade durability with a setup measured in seconds, not half a day.