I understand the intent is to remove select columns, then the simple DELETE would do.
The other way of doing things is probably having empty placeholders for the dates/data. /Arthur From: Suruchi Deodhar Sent: Friday, August 16, 2013 2:38 PM To: [email protected] ; Arthur Zubarev Subject: Re: Deleting column data from Cassandra without setting its TTL We are using the cassandra thrift interface to interact with Cassandra. In our use case, we store time-series data with one column per day (per row-key) , and we have a fixed number of column families to store data over time. Since we have data for hundreds of days, we cannot create column-families for individual time-ranges easily ( if that's what you're suggesting). Is there another way we can achieve what we intend to do ? Thanks On Fri, Aug 16, 2013 at 2:29 PM, Arthur Zubarev <[email protected]> wrote: Not sure what client you use to interact with C*. In your scenario I assume it is CQL. So what can go wrong with TRUNCATE? Regards Arthur From: Suruchi Deodhar Sent: Friday, August 16, 2013 12:23 PM To: [email protected] Subject: Deleting column data from Cassandra without setting its TTL What is the best way to delete column data from a Cassandra cluster after it is backed up with snapshots (using Priam or otherwise) to a stable storage? We dont want to use the TTL option, since we would like to restore the backed-up data in the Cassandra cluster at a later time. Setting a TTL would enable the tombstone marker, which would make it impossible to restore data. What are the available options in Cassandra 1.2.5 in particular? Thanks!
