Build Anything with FoundationDB
Talkat FoundationDB Summit 2019 · on Monday, 2019-11-18· at 16:00· Grand Ballroom 5, Marriott Marquis San Diego Marina·
summary
Starts from an empty key value store and shows how to model a relational schema using the tuple layer, primary and secondary indexes that store the primary key for range scans, and a background backfill that finishes an index while avoiding false negatives. It uses a metadata version key and non-blocking schema changes via atomic metadata updates, version‑stamped writes for a change data capture log sharded across partitions, and a pattern that achieves consistent object listing for S3 by writing objects to S3 then atomically storing pointers in FoundationDB with eventual GC.
AI-written summary · about
chapters (22)
- 0:00Intro
- 1:17Database Construction
- 2:44Detour: CAP
- 4:47Consistency using Foundation DB
- 5:27(Relational) Schema Objects
- 6:29Tuple Encoding CREATE TABLE Students
- 7:39Index Tuples
- 8:50Index Key Tuples
- 11:25Non-blocking Schema Changes
- 12:06Metadata Version Feature
- 12:59Online, Asynchronous Schema Change in F1
- 13:38Adding an Index
- 15:37Change Data Capture
- 16:31Implementing CDC in FDB
- 17:21What are "old" and "new"?
- 18:31CDC Key Structure
- 19:23Feature Request
- 20:32Amazon S3 Limitations
- 21:55So what can S3 do?
- 22:26How to fix S3: Listing
- 23:03How to fix S3: Put
- 24:12How to fix S3: Delete
related talks
37mTechnical Overview of FoundationDB
22mSolving Everyday Data Problems with FoundationDB
46mFoundationDB: The Best-Kept Secret of New Distributed Architectures!
50mBuilding the Next-Generation Backend on FoundationDB
22mHow FoundationDB powers SnowflakeDB's metadata!
53mBuilding Distributed, Multi-Tenant Databases with FoundationDB and Rust
34mFoundationDB, the Black Knight
26mFuture of FoundationDB Storage Engines