Based on
http://docs.datastax.com/en/cassandra/2.0/cassandra/dml/dml_config_consistency_c.html
Write Consistency
The consistency level determines the number of replicas on which the write must succeed before returning an acknowledgment to the client application
Read Consistency:
The consistency level specifies how many replicas must respond to a read request before returning data to the client application.
This means, if I specify CL as Quorum for both read and writes, within the same client application (and presumably the same thread), my consequent read of the data (with CL = Quorum) after a write (again, with CL = Quorum) ensures I get the latest data.
What I am not very clear about however is: does this apply across different processes and different threads within the same application as well?
Is it possible to specify CL = Quorum and yet have missed updates (old data) or dirty reads?
Update:
To clarify, I'm not talking about AFTER the CL QUORUM has finished. The question is more along these lines:
If in the middle of the write (@ QUORUM) another process connecting to the same cluster attempts to read the same partition with CL = QUORUM, would it be possible to read old data?
Imagine the write hasn't yet finished writing to all the replicas so the CL hasn't yet been satisfied and a read occurs. Could it be possible that the write has only been written to one node (and in the process of being replicated) and therefore, when the read (with CL = QUORUM) comes in, it reads the older data? Do the nodes somehow know they should "wait" until the writes have finished
Many thanks,
