Posted: 26 May 2018 9:22 EDT Last activity: 30 May 2018 20:14 EDT
Storing Interaction History in Cassandra/Hadoop
As the Interaction History increase by volume very quickly and also demand very fast access for different decsioning and batch processing, is it possible to keep Interaction History in the Cassandra (Decision Data Store) data set?
Or looking at the complex relational structure of different Interaction History related Fact and Dimensions tables, it is not feasible.
If you think storing subset of IH records in DDS helps in your use-case. Then you may do so by seperating out your actual Data Flow run, so that source for your main DF run can be initialized by feeding the DDS Data Set from IH.
> is it possible to keep Interaction History in the Cassandra (Decision Data Store)
We don't support that. IH currently is relational.
Note that this is not because of the complex structure. You can store any complex class in DDS.
What are you using IH for? What kind of very high volume data are you storing in it that you raise this as a question? Note that by no means IH is the only way to store interaction data - you can store it outside of IH in DDS or other ways if you want to.
the requirement is for outbound marketing campaign among millions of customers and over different channel (Client's website, paid channel,) and we want to capture click trough rate in our Central Customer Decision Hub.
so was thinking whether it will be any performance issue if Relation database based IH is used for capturing the customer response which will be high in volume and velocity.