not a support question
thomical
Member
Hi,
With support for composite row keys, I am contemplating using a composite primary key to manage time series data and keep the columns "thin". This is to allow for Hive integration with my timeseries CF. The row key would look like this:
<date_stamp>:<event_id>
If I understand this correctly, my data would be bucketed by day and partitioned by day on the same nodes, hence I wouldn't get a performance penalty.
Is this assumption correct?
Thomas
Thomas,
Are you referring to using a composite in your row key or referring to the compound primary key support in CQL3 described here:
http://www.datastax.com/dev/blog/whats-new-in-cql-3-0
I'll assume you are referring to manually created composite row keys and not cql3. In that case each row key of <date_stamp>:<event_id>, would exist on a different node in the cluster. So if you are new time based columns for each event_id then, yes your columns for each event_id would be bucketed by day and exist on a single node.
I recommend reading http://www.datastax.com/dev/blog/advanced-time-series-with-cassandra
You must log in to post.
