I’m intrigued by the blocking skip list throughput numbers. By description (and assumption) I would expect it only to block for writes, but the throughput stays static even as the read/write mix changes. Is this expected?
There are other memtable implementations coming too, I think? It looks like 7282 is ready to rock once this lands, and perhaps has even been prepared also by Branimir (I recall seeing it in an earlier PR)? Might be nice to call out another implementation. (I expect at least one more major implementation, but that’s probably a ways off and I don’t think has any suitable ticket or documentation to link) It looks like a great post though. > On 14 Jul 2022, at 19:52, Henrik Ingo <henrik.i...@datastax.com> wrote: > > > Thanks. This is concise and well written, yet exciting already to see the > sharded skip list results, and with exciting "more to come" at the end. This > is a big deal - I believe - for Cassandra and therefore I appreciate that > this blog post will be published. > > henrik > >> On Wed, Jul 13, 2022 at 6:36 PM Chris Thornett <ch...@constantia.io> wrote: >> The 72-hour window for community review is now open for Branimir Lambov's >> blog on Pluggable Memtable Implementations. Please indicate any amends in >> the comments. Thanks! >> https://docs.google.com/document/d/1Iws9gQZCp_80Be_ZfuyAB82X1EhqWihEb35Sevv-JDE/edit >> -- >> >> Chris Thornett > > > -- > Henrik Ingo > +358 40 569 7354 >
