C code - not only code but also documentation - RESP3 protocol - compare it with YAML spec - TCP - network is the bottleneck (or not?) - It’s everywhere - see https://db-engines.com/
C code - not only code but also documentation - RESP3 protocol - compare it with YAML spec - TCP - network is the bottleneck (or not?) - It’s everywhere - see https://db-engines.com/ - PERFORMANCE!!11!!11
- create commands -> submit to client -> get response - Returning ‘interface{}’ (or a modern ‘any’) allocates - which you cannot omit - Not all clients understand context package - this can be achieved via wrapper or timeouts Observations 38
- create commands -> submit to client -> get response - Returning ‘interface{}’ (or a modern ‘any’) allocates - which you cannot omit - Not all clients understand context package - this can be achieved via wrapper or timeouts - Reusing memory is completely impossible - Go GC is a cool thing but we can do better Observations 39
and stable - It’s always YMMV - your mileage may vary BTW: this slide must be in every presentation where “benchmark” is mentioned ⚠ Benchmark disclaimer 58
19964.06 requests per second > redis-benchmark -n 100000 -P 64 set key value Summary: throughput summary: 446571.41 requests per second -P = Pipeline <numreq> requests. Default 1 (no pipeline). Redis bench 61