Comments by "Mikko Rantalainen" (@MikkoRantalainen) on "The FASTEST and SAFEST Database" video.

  1. You can also do general purpose deterministic fault checking with open source LFI toolkit but I'm pretty sure that way would be much slower to generate lots of test cases than running the code in simulator as explained in this talk. Another option is libfiu.
    2
  2. Deterministic simulation with single random seed was pretty cool idea.
    2
  3. If you had so many transactions that the single CPU thread on current database master machine cannot handle it, what then?
    1
  4. Interesting architecture. I would have guessed that having constant wallclock timing for "airplanes" would have been even better than having a constant limit of "passangers" per plane. For example, send new airplane every 10 ms and you'll get 10 ms latency for any given transaction as long as you can handle any plane in 10 ms.
    1
  5. Starting with P100 latency when you start improving the performance of the whole system is a great starting point of any system. It's still surprising that Zig was selected over Rust. I understand that Zig implementation was probably easier to implement but Rust implementation would have provided more safety against implementation errors which I would have considered critical for financial database. If your runtime safety depends on additional runtime sanity checks, are you really faster than Rust implementation that can guarantee the same safety with compile time checks? Of course, if you want to catch CPU errors, too, you obviously need also runtime checks. I think financial database should have all logic checked at compile time and also have runtime checks to verify the hardware.
    1
  6.  @drofdissonance  Full control over allocated memory could definitely be one valid reason. That was the main pain point of Rust that had to be fixed before Rust was allowed in Linux kernel, too.
    1