Wild ass guess based on what worked for my previous employer (for whom I wrote twcs), but note that if you model it such that partitions are contained to a single window/sstable that number is probably irrelevant and may be able to be made arbitrarily high
> On Aug 21, 2019, at 4:48 PM, Oleksandr Shulgin <oleksandr.shul...@zalando.de> > wrote: > > Hi, > > In the TWCS readme[1] it says: > > You should target fewer than 50 buckets per table based on your TTL. > > What is this recommendation based upon? The docs from Datastax[2], for > example do not mention anything like that. > > Is the recommendation still actual today for the version 3.0.$latest? What > is the computation complexity behind it: is it N, N*log(N) or N^2, etc.? > > Thank you. > > [1] https://github.com/jeffjirsa/twcs > [2] > https://docs.datastax.com/en/archived/cassandra/3.0/cassandra/dml/dmlHowDataMaintain.html#dmlHowDataMaintain__twcs > > -- > Alex >
