We opted for a no SQL kind of data storage. We looked at Redis and MongoDB and opted for Couchbase because it had SQL integrations, programming language and scripting language.

We were able to evaluate Cassandra. When comparing the two: Couchbase is more general purpose, and it has a smaller latency, while Cassandra is easier to manage with the open source version in cluster environments.

We evaluated other options as well, but we choose Couchbase for the immediate consistency with Elastic, and the real-time native replication plugin, which is fantastic!

We also looked at MongoDB, some sort of eMemory, and cache layered with a traditional relational database for performance sake, because we churn a lot of data. We collect for each customer upwards of around 20GB/hour of data. Therefore, performance is important to us, so NoSQL eventually became the obvious choice.

The decision was between this solution and MongoDB. We chose this because it came with N1QL, which is their query language. Whereas MongoDB, at that time, did not come with any sort of query capability.

I have evaluated MongoDB and it is lacking some features that Couchbase has making it a complete solution. With MongoDB, we would need another solution to work alongside it to function the same as Couchbase.