Re: [ANN] Khepri 0.2.1 - A tree-like replicated on-disk database library
Karl Nilsson <[email protected]> Thu, 3 Mar 2022 09:53:53 +0000
| Newsgroups | gmane.comp.lang.erlang.general |
|---|---|
| Message-ID | <CAHC35TBR7ZYHhm0v1DcGcQHh6+eCSHSCH-xVJp8b6bbKwbP0SQ@mail.gmail.com> |
Frank those figures on ci are quite different from what we saw when run on more powerful systems. It would be great if you could run them yourself and share the results. On Wed, 2 Mar 2022 at 23:13, Frank Muller <[email protected]> wrote: > Hi Jean-Sébastien, > > Very promising. Would it be possible to test insert (50%) + delete (25%) + > update (%25) in the same time? You can vary the numbers of course. > > How about “gets”? Are they faster in Khepri? > > Best > F. > > [email protected]> a écrit : > >> On 18/02/2022 11:21, Frank Muller wrote: >> > Do you provide (throughput / latency) benchmarks? >> >> Hi! >> >> I prepared a first micro-benchmark tool to compare Khepri and Mnesia: >> https://github.com/rabbitmq/khepri-benchmark >> >> The README explains what it measures and how it works. In particular, it >> only tests operations per second, not latency. >> >> Results are automatically published from GitHub Actions to the following >> page (though remember that the CI runner only has 2 CPU cores): >> >> >> -- >> Jean-Sébastien Pédron >> > -- *Karl Nilsson*