Open-sourcing a 10x reduction in Apache Cassandra tail latency

At Instagram, we have one of the world’s largest deployments of the Apache Cassandra database. We began using Cassandra in 2012 to replace Redis and support product use cases like fraud detection, Feed, and the Direct inbox. At first we ran Cassandra clusters in an AWS environment, but migrated them over to Facebook’s infrastructure when the rest of Instagram moved. We’ve had a really good experie

[rjj]

"After investigation, we found the JVM garbage collector (GC) contributed a lot to the latency spikes." "the P99 read latency dropped from 60ms to 20ms. We also observed that the GC stalls on that cluster dropped from 2.5% to 0.3%"

[daisukebe]

強い

[K2ICE]

“we decided to build the new storage engine on top of RocksDB. ”

[tarhashi]

Rocksandraだと？

[ono_matope]

Rocks! もう分散システムはJavaであることが最大のボトルネックみたいな感じだ…

[YuichiTanaka]

CassandraがGCでストールしちゃうので、RocksDBを組み込んで、レスポンスタイムのp99が60msから20ms、GCストール時間の割合も2.5%から0.3%に改善したという話。強すぎ。

[yosida95]

カッコいい

[LaclefYoshi]

RocksDBを組み込んで高速化