On Sat, Oct 13, 2012 at 6:27 AM, Richard Tang <[email protected]> wrote: > Hi, I want to manually control the number of versions of data physically > stored in hbase. > > I am aware that physically deleting a record in hbase occurs only in major > compaction > (link<http://hadoop-hbase.blogspot.com/2011/12/deletion-in-hbase.html>), > and want to know if it's possible to physically delete only the records in > major compaction, that meets, for example, the following two conditions: > 1. has more than one version. (>=2) > 2. timestamp is lower than a specified value. > > Does hbase provide interface for config compaction to allow the above > behaviour (without change hbase code and recompiling)? Let me know if you > have any thoughts on this. Thanks. >
Take a look at coprocessers. You can interpolate your own code at compaction time IIRC. St.Ack
