hn.today

Jepsen: ParticleDB 1.1.4

jepsen.io146 points75 comments
Screenshot of Jepsen: ParticleDB 1.1.4

Jepsen tested ParticleDB, a database intended to combine SQL, key-value storage, vector search, analytics, and programmable functions for AI and workflow applications. The review examined versions 1.1.4 through 1.1.18.11.rc-22 on single nodes, since the supplied builds did not support the product’s replicated mode. ParticleDB advertised “Full ACID,” but its documentation gave conflicting descriptions of transaction isolation and durability, and different interfaces disagreed about which settings were defaults or what they guaranteed. Jepsen tested transaction modes and WAL sync options using concurrent reads and writes, appends, crash and disk-failure simulations, and generated SQL comparisons against PostgreSQL.

Testing uncovered 96 issues, including crashes, deadlocks, lost data, inconsistent results within a transaction, malformed or reappearing tables, and values or columns changing unexpectedly. Many defects required no failures or complex setup - just one server and client. Most were apparently fixed by version 1.1.18.11.rc-22, but important safety failures remained: the default OCC mode could lose committed writes during ordinary operation, while the strongest serializable setting still exhibited internal anomalies and Lost Update. The findings therefore challenge ParticleDB’s claims of full transactional safety, not just its reliability under crashes. Orkes later released version 1.1.19-rc.13 and said it fixed all remaining issues in the report; the testing was funded by Orkes.

Read on jepsen.io75 comments on Hacker News

Summary generated by AI from the linked article. hn.today is not affiliated with Hacker News or Y Combinator.

More in Programming

Shaun Thomas: Burnt Toast

Shaun Thomas: Burnt Toast

PostgreSQL’s TOAST storage made the table much smaller, and counting or scanning rows was faster than with the larger uncompressed table. But reading a field inside every toasted value required far more buffer reads, showing that TOAST can trade space savings for slower access to large values. (pgedge.com)

The daily digest

Today's best Hacker News stories, summarized and screenshotted, one email a day.