Well, my concern is that when data is stored internally in HFiles, each key belongs exactly to one file.
So I'm wondering if HBase just adopts HFile during import or tries to merge it with others On Mon, Oct 24, 2011 at 10:17 PM, Ravikumar MAV <[email protected]> wrote: > New row many have new version(ts) for each column value (unless table is > configured to have one version or both rows are generated with same version > ts). Atleast tsvimport generates new ts for this situation. > > > On Mon, Oct 24, 2011 at 10:53 AM, Vladimir Klimontovich < > [email protected]> wrote: > >> Hello! >> >> Could please someone suggest how HBase behaves in following situation: >> >> - I have a row with key K (row1) that already present in table. >> - I'm running a mapreduce job which generates HFile that contains row >> with the same key K (row2) >> - Will the data of this rows (row1 and row2) will be merged somehow >> or row2 will completely replace row1 >> - Almost the same case, but mapreduce job outputs too HFiles and each >> of them contains rows with same key K. Will rows be merged? >> >> Thanks! >> >> >> -- >> Vladimir Klimontovich >> Cell: +7-926-890-2349, skype: klimontovich >> > > > > -- > - Ravi MAV > -- Vladimir Klimontovich Cell: +7-926-890-2349, skype: klimontovich
