Measurement records
Each file is one dated measurement: the machine, the build, the method and the numbers as they were recorded that day. A record is not updated afterwards, so its commit, its numbers and its “open” notes describe that date only. A new measurement goes in a new file named YYYY-MM-DD-topic.md. Rules for measuring are in PERFORMANCE and BENCHMARKING.
Current numbers are on the performance and memory pages.
| Date | Record | What it measured |
|---|---|---|
| 2026-10-16 | Eight shards against Redis | 60 commands and 12 data structures, Ruvio with eight shards against one Redis core and an eight-node Redis Cluster on one Azure EPYC 9V74 VM; one-shard control run |
| 2026-10-11 | Azure command comparison | 60 commands and 12 data structures against Redis 8.10.2, throughput and memory, one Azure EPYC VM |
| 2026-10-01 to 2026-10-07 | Benchmark results | Native Redis comparison and native JSON vs RedisJSON on an Apple M4 Pro, capped Docker Desktop comparison, hash-field TTL, sorted-set pop, SEMAPHORE and INCRBY MAX commands |
| 2026-10-03 | .NET scenario lab | k6 load on a .NET API using Ruvio or Redis through the C# client, persistence on and off, client multiplexing |
| 2026-09-28 | Bare metal | Cherry EPYC 7313P Linux host: matrix against Redis, durability profiles, 15-minute soaks, kill -9 trials |
| 2026-09-28 | String memory density | Where a string key paid memory in the Cherry soak, and the order of work that followed |