All, Thank you for giving lots of opinions and information. I'll try to persuade my colleagues as follows:
I couldn't find any good examples where versioning should be definitely utilized. However, HBase community members gave me the idea on how versioning is useful. 1. Recover data lost by accidental deletions or updates (I think this is the most persuading reason) 2. Auditing (change tracking) )for compliance However, this is not persuading, because advanced RDBMSs provide audit trails, not versioning. Versioning itself does not show who changed the data how. 3. Recording events (as in Google's persolalized search) This is not persuading, too. As I wrote in the previous mail, embedding time of event in row key may be better because it prevent the rows from becoming big. If versioning is not necessary from your requirement, you can ignore timestamps (do not have to specify timestamp in API call). Although HBase keeps three versions by default and it may be a bit wasteful for memory and disk, turning on compression for column families can minimize the waste as much as you can ignore (is it true?). If saving memory (=keep memtable as small as possible) is important, you can set the maximum number of versions to 1. The reason that the default is 3 is to rescue users from their mistakes. (If users accidentally delete or update data, you have to develop a tool that pulls previous data records.) Regards Takayuki
