What a NoSQL ... again :) but a different use case

Suggestions for a NoSQL data warehouse so that we can easily transfer data and generate reports in real time Qlikview?

Easily means: 1. Qlikview read support ( connector available, otherwise it could possibly write a JDBC connector, otherwise it could possibly write a custom QVX connector in the datastore)

  • Easily adapts to schema changes or schematization. We change our schema often ...
  • Java support for recording
  • Super Fast View - real-time incremental access as well as batch access for old data within the time range. I read that Cassandra excels in ranges.
  • Fast enough recording
  • Large enough data store - 20 million rows stored per day, about 200 bytes each
  • It would be nice if it could scale over several years, and elasticity is not that important.
  • Easy to use, install and run. Looking at the minimum setup and tuning times.
  • Matlabe support for adhoc queries

At first I don't think we need a distributed system, but a cluster is an opportunity.

I looked at Mondobba, Kassandra and Hbase. I don't think moving to REST is a good idea because of the (theoretically) slow performance.

I'm leaning towards MongoDB for now because of its ease of use, Matlab support, schema is completely smaller, Qlikview support (beta connector available). However, if anyone can suggest something better that would be great!


source to share

1 answer

Depending on the server infrastructure used, I find the amazon NoSQL service available at aws.amazon.com to be the best choice .

The point is that any DB will have poor performance in the cloud infrastructure due to the way it stores data, for example Amazon EC2 with EBS, VERY slow for this task, requiring up to 20 EBS volumes connected in a raid to acquire decent speed. They solved this problem by creating this NoSQL service that I have never used, but I find it nice.



All Articles