+1 (nb) - wiłl be nice for the analytics writer to be able to size SSTables appropriately and efficiently.
Doug > On Oct 24, 2023, at 10:36 PM, guo Maxwell <cclive1...@gmail.com> wrote: > > 😄 > > Chris Lohfink <clohfin...@gmail.com <mailto:clohfin...@gmail.com>> > 于2023年10月25日周三 05:02写道： >> +1 >> >> On Tue, Oct 24, 2023 at 11:24 AM Brandon Williams <dri...@gmail.com >> <mailto:dri...@gmail.com>> wrote: >>> +1 >>> >>> Kind Regards, >>> Brandon >>> >>> On Mon, Oct 23, 2023 at 6:22 PM Yifan Cai <yc25c...@gmail.com >>> <mailto:yc25c...@gmail.com>> wrote: >>> > >>> > Hi, >>> > >>> > I want to propose merging the patch in CASSANDRA-18941 to 4.0 and up to >>> > trunk and hope we are all OK with it. >>> > >>> > In CASSANDRA-18941, I am adding the capability to produce size-bounded >>> > SSTables in CQLSSTableWriter for sorted data. It can greatly benefit >>> > Cassandra Analytics (https://github.com/apache/cassandra-analytics) for >>> > bulk writing SSTables, since it avoids buffering and sorting on flush, >>> > given the data source is sorted already in the bulk write process. >>> > Cassandra Analytics supports Cassandra 4.0 and depends on the >>> > cassandra-all 4.0.x library. Therefore, we are mostly interested in using >>> > the new capability in 4.0. >>> > >>> > CQLSSTableWriter is only used in offline tools and never in the code path >>> > of Cassandra server. >>> > >>> > Any objections to merging the patch to 4.0 and up to trunk? >>> > >>> > - Yifan
