That would depend on how much data is generated per day. If it can still fit in a row, the solution wold be to to just have rows keyed by date, like 20110326. This way you don't have to move data inside the cluster, the selection logic will be in the client.
Even if the data is too large to be put in a row for single day, you can store all IDs of objects in this fashion, that's usually practical (plus you can make use of column names for additional indexing capability). Then, you have to have a separate (and potentially very large table) which contained the actual data. Essentially, one table would serve as an index to the other, which is the data store. I have a working application that's structured like that and it works OK. -- View this message in context: http://cassandra-user-incubator-apache-org.3065146.n2.nabble.com/help-modeling-a-requirement-in-cassandra-tp6209726p6210994.html Sent from the cassandra-u...@incubator.apache.org mailing list archive at Nabble.com.
