Jingyu Zhou·Sep 21Gray Failure, Fail-Silent Failure, and Metastable Failure in FoundationDBDistributed systems do not fail in only one way. A process can stop completely, a network connection can become slow, or a component can…
Jingyu Zhou·May 31Consistency Checkers in FoundationDBFor any distributed database, data consistency is one of the most critical objectives. The need for multiple replicas to provide high…
Jingyu Zhou·May 17Success Is Its Own PerilFoundationDB’s obsession with correctness has created an unexpected blind spot: poor tooling for backup recovery.A response icon1A response icon1
Jingyu Zhou·Apr 26Understanding FoundationDB’s causal_read_riskyFoundationDB has a transaction option (and a database option transaction_causal_read_risky that forwards the same setting to every…
Jingyu Zhou·Apr 18Redundancy and Fail-silent: Lessons from NASA’s Artemis IIThe Story of Artemis II’s Fault-Tolerant Computer
Jingyu Zhou·Apr 11Modeling Commit Latency in FoundationDBWhen a transaction commits in FoundationDB, the commit proxy fans its log records out to a set of transaction log nodes (tlogs), waits for…
Jingyu Zhou·Feb 21Safe Storage Engine Testing in ProductionPreviously, I’ve discussed Why We Migrated FoundationDB Clusters from SQLite to RocksDB Storage Engine. What I have not discussed in that…
