Consistent Caching in FoundationDB
Talkat FoundationDB Summit 2019 · on Monday, 2019-11-18· at 13:00· Grand Ballroom 5, Marriott Marquis San Diego Marina·
summary
Proposes a native cache that holds hot key ranges in memory and increases replication factor for those ranges to throw more CPUs at load; introduces ephemeral, stateless storage cache servers that pull mutations from TLogs, apply them, and serve reads while writes still go to TLogs. Clients keep metadata to route reads to caches or storage servers and update it when servers report cached ranges. Automatic detection uses per-storage-server request stats plus a central coordinator; manual mode lets admins pin any range.
AI-written summary · about
chapters (11)
- 0:00Intro
- 0:20Motivation
- 2:11Introduction
- 3:03Proposed Solution
- 4:09Block Diagram
- 5:24Storage Cashflow
- 6:20Client Side Changes
- 7:45Implications
- 8:10Configuration Modes
- 9:08Automatic Detection Catch Management
- 10:41Conclusion
related talks
37mTechnical Overview of FoundationDB
22mSolving Everyday Data Problems with FoundationDB
19mCross-Region Replication Options with FoundationDB
26mFuture of FoundationDB Storage Engines
46mFoundationDB: The Best-Kept Secret of New Distributed Architectures!
53mLeveraging Caching For Cloud Native Applications
50mLet's talk about the cache!
34mFoundationDB, the Black Knight