Hi Tuco, Backup partition is full copy of primary partition, so you will run sql query on same data. There is no need to run query on backups as we can run in on primaries in multiple thread with same effect.
We already test approach with running query in multiple threads when every thread works with its own partition set user filter. It get us much poor throughput scalability with linear CPU pressure than approach with segmented index. On Tue, Apr 25, 2017 at 12:18 PM, tuco.ramirez <[email protected]> wrote: > If SQL queries would execute not only on all masters, but on at least one > instances from (M1,S11,S12)+(M2,S21,S22)+(M3,S31,S32), then the reads > could > be scaled linearly by increasing the no of slaves. > ie a query would execute not on all masters, but on any combination of > nodes > where complete data was present. > Is it planned in the foreseeable future? > Thanks > Tuco > > > > -- > View this message in context: http://apache-ignite-users. > 70518.x6.nabble.com/Basic-question-on-scaling-of-reads- > with-ignite-sql-queries-tp11975p12226.html > Sent from the Apache Ignite Users mailing list archive at Nabble.com. > -- Best regards, Andrey V. Mashenkov
