Environment: Jruby, Rails 2.3.8, cassandra-gem, cassandra 1.1
We are creating a new keyspace per tenant in our cassandra backed multi-tenant application. While testing we found Cassandra fail
at 400 keyspaces on a 8GB RAM machine, 7200 RPM disk
at 700 keyspaces on a 24 GB RAM machine, 7200 RPM disk
Also found it run painfully slow after 100 or so keyspaces. It took ~7 hours to create 700 keyspaces and consumed ~35GB of disk space with no data entered into it.
We then switched our testing to validate the maximum number of column families in a keyspace, hoping to use <tenant id>_<columnfamily name> named CFs. This test also failed at ~3000 CFs.
Now we are looking at <tenant_id>_<key> for row keys and use the same CFs for all tenants. This is based on the notes at https://github.com/rantav/hector/wiki/Virtual-Keyspaces.
The question is, would prepending tenant_id to key work for the following CF, given the key validation class is LongType?.
ColumnFamily: monthly_unique_user_counts
"count of unique users per month"
Key Validation Class: org.apache.cassandra.db.marshal.LongType
Default column value validator: org.apache.cassandra.db.marshal.LongType
Columns sorted by: org.apache.cassandra.db.marshal.LongType
Row cache size / save period in seconds / keys to save : 0.0/0/all
Row Cache Provider: org.apache.cassandra.cache.ConcurrentLinkedHashCacheProvider
Key cache size / save period in seconds: 200000.0/14400
For a few other CFs, the Key Validation Class is
Key Validation Class: org.apache.cassandra.db.marshal.TimeUUIDType
Would the <tenant_id>_<key> concept work in that CF?.
