Hi Manhua, Page size will be smaller/Equal to 32k but here point is how much IO benefit will get during query. Since you are adding bloom information inside page metadata, it might benefit in processing time with little bit of IO overhead. So its better to keep Bloom information at blocklet level in footer.
-Regards Kumar Vishal On Thu, Nov 28, 2019 at 8:21 PM Jacky Li <[email protected]> wrote: > Hi, > > Since the new bloom filter integration is designed to work in the executor > side, so in term of keeping it simple, I actually prefer to keep it inside > the data file itself instead of keep in a separated index file. So in > executor side, only read one file is enough. And it is always better if we > can follow the datamap interface. > > Regards, > Jacky > > > > -- > Sent from: > http://apache-carbondata-dev-mailing-list-archive.1130556.n5.nabble.com/ >
