Redis

Redis is an open source (BSD licensed), in-memory data structure store, used as a database, cache and message broker. It supports data structures such as strings, hashes, lists, sets, sorted sets with range queries, bitmaps, hyperloglogs, geospatial indexes with radius queries and streams. Redis has built-in replication, Lua scripting, LRU eviction, transactions and different levels of on-disk persistence, and provides high availability via Redis Sentinel and automatic partitioning with Redis Cluster.

Optimization

How Fast Can A Single Redis Be

• Used a multi-threaded fork of Redis called KeyDB
• Default execution behavior is to send syscalls to the kernel
  • Bottleneck becomes processing the data packets
  • Solution is to move execution to user space via Intel’s DPDK
• Able to increase perf over Redis 5.0
  • 164,000 ops/sec -> 300,000 ops/sec
  • 1.8x latency reduction
• Tested with Memteir running on a m5.4xlarge machine with Redis running on a m5.2xlarge machine

Before

After

Client Side Caching

Original Post Original Proposal

• Every read-only command will return the keys and keep track of the client that requested the key
• Invalidations are done on key slots
  • Multiple keys could get invalidated at the same time (though the number of keys invalidated is small, 3-4)
• Invalidations are also done when a client is disconnected
• Part of the new RESP3 protocol in Redis 6 (as of 2019/07/10 in unstable)
• Enabled in Redis 6 by opting into the feature: CLIENT TRACKING on