This is indeed a question about Orient, but first a bit of background...

We have several microservices utilizing Cassandra for persistence. Until 
now, our use cases have been pretty tolerant of eventual consistency and 
the BigTable model. However, now we need to model simple relationships 
between users and userGroups in hierarchies, and certain correctness checks 
like cycle detection come into play. We can make it work on Cassandra with 
a few cheats: 

   - Store entire graph under a single Cassandra row and use Paxos to force 
   lineralizable consistency
   - Skip correctness checks on writes, only write mutations, resolve the 
   final state on reads and somehow resolve or handle invalid mutations on 
   read as well.
   - Slap Titan in front of Cassandra and be ready to handle invalid states 
   on reads

Essentially, this all fits in well with CAP as I can either kill 
availability for consistency or vice-versa. If my team were the only one 
that needed to do this, then we would just deal with it, but we are also 
tasked with establishing patterns for other teams in the org to use - 
"figure out the hard stuff first and streamline it for the others".

So I find OrientDB, with multi-master replication AND acid? I've done a 
little bit of reading on Hazelcast and Orient, but I'm having difficulty 
sorting out the marketing from the reality.

The question...

Can someone provide a comparison between the consistency characteristics of 
Cassandra and OrientDB. I have a good grasp on the pitfalls and limitations 
of Cassandra, but I would like to understand the same for Orient.

Thanks!

-- 

--- 
You received this message because you are subscribed to the Google Groups 
"OrientDB" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to [email protected].
For more options, visit https://groups.google.com/d/optout.

Reply via email to