implement a multiput function around the existing put function that flushes at the end.
Set and classify
Is this not simply achieved by setting flush/buffersize hints to the store? For example if you set autoflush to false and possibly increase the flush buffersize then you effectively already have a multiput, no?
Yes, I had not checked the autoflush/buffersize option. Thank you
Ok. Please note that deletes are currently still pretty inefficient to process. (Deletes are always processed immediately regardless of autoflush. After I had done some tests with the HTable supported batch deletes, this seem to be slow too). It's something to keep in mind. HBASE-3584 will probably fix this.