I would think that a single BM ought to perform well over time, although I'd be interested in hearing otherwise.
FWIW, BM's criteria for sending batch requests is more about size of the puts rather than the number of puts. On Wed, Oct 28, 2015 at 6:17 PM, Artem Ervits <[email protected]> wrote: > Thanks Enis, would you say creating a new BM every 10k rows vs reusing the > same BM would make any difference in performance? > On Oct 28, 2015 2:54 PM, "Enis Söztutar" <[email protected]> wrote: > > > Hi Artem, > > > > The 1.0 API BufferedMutator should cover the use case where previously > > HTableUtil was used or HTable.setAutoFlush(false) is used. > > > > BufferedMutator already groups the mutations per region server under the > > hood (AsyncProcess) and sends the buffered mutations in the background. > > There should be enough documentation to get started, and there is example > > code in hbase-examples module for BM usage. But let us know if it covers > > your use case or not. > > > > Enis > > > > On Tue, Oct 27, 2015 at 7:58 PM, Artem Ervits <[email protected]> > > wrote: > > > > > Hello all, > > > > > > is there an equivalent utility in HBase 1.0+ to HTableUtil? I'd like to > > > batch mutations by region server. > > > > > > Thanks > > > > > >
