[ https://issues.apache.org/jira/browse/HBASE-10115?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13844383#comment-13844383 ]
Jean-Marc Spaggiari commented on HBASE-10115: --------------------------------------------- I had few usecases in mind when I wrote that. 1) You have one table with some events/data/information with a VERSION=5 limit because you want to keep if "small" and efficient. However, when a version is removed, you don't want to loose it because you need it for other purposes. So you want to "transfer" it to another table where you will have VERSION=MaxInt. 2) You have a TTL of 30 days in a column because you only need quick access to the last 30 days values. However, you want to keep he full history in HDFS. So each time a value expire from the table, you want to be able to store it into HDFS. You can steam it to flume, or push it on a queue, or write it into HDFS then aggregate the values, or anything else. 3) You have some mandatory requirements to keep EVERYTHING. Even what has been deleted. So when a delete is applied while compacting, you want to store somewhere else what has been deleted, just in case. Like a ".trash" option, but for cells. There might be others, but that was all I had in mind at that time... > Add a hook for compaction deletes > --------------------------------- > > Key: HBASE-10115 > URL: https://issues.apache.org/jira/browse/HBASE-10115 > Project: HBase > Issue Type: Bug > Components: Compaction, Coprocessors > Affects Versions: 0.99.0 > Reporter: Jean-Marc Spaggiari > Priority: Minor > > The idea is to add a hook to coprocessors to be able to get triggered when > compaction delete a cell before of the number of versions or the TTL. -- This message was sent by Atlassian JIRA (v6.1.4#6159)