1. 35

  2. 14

    VoltDB doesn’t have a whole lot to sell to me, really. That said, if I understand the chain of events properly, VoltDB sponsored this Jepsen post (and all the research that went into it), took those findings and started fixing the problems unearthed. That’s an admirable commitment to both data safety and openness, and it means I’ll consider VoltDB preferably over competitors should I ever need that feature set.

    1. 28

      Yep, you’re understanding correctly. Like RethinkDB, VoltDB approached me for help in testing their systems, and funded the research. I found initial cases pretty quickly, deeper problems over the next month, and worked with their team for the next month or so to create more stringent tests and evaluate proposed fixes–VoltDB ported some of these test scenarios to their internal test suite, and is integrating Jepsen into their testing cycle now. That work culminated in the release of 6.4 last week. You can read more about how I handle sponsored research, and see the full set of bugs we uncovered on VoltDB’s issue tracker.

    2. 3

      I don’t work with ditributed databases, but always find these Jepsen posts fascinating. It’s so interesting to read how a true expert evaluates a system. The fact that the report is published for everyone to see, makes it even more compelling – there is no hiding from the results.

      I’d be terrified to have my software tested this way.

      1. 4

        Speaking from experience, it’s quite terrifying. But as you think about it, there are few bad outcomes possible, so long as you take any found issues seriously and work to fix them ASAP.

        1. 4

          wouldn’t you be more terrified that you’re letting your customers down if you did have these issues and did not find and address them?