Closed dgryski closed 9 years ago
Fixed in https://github.com/dgryski/carbonmem/commit/a243eabb3ebf6220eeba812c5778339cdc1a4294 . Speedup wasn't quite 60x as each minute now contains more metrics to scan through. However, sharded metric spaces (issue #2 ) and this both combined to speed up queries nicely.
Allow rollup periods to reduce memory usage.
In addition to memory reduction, the minutely data can also be used for TopK queries, reducing work by ~60x.