Saiku Analytics vs. Pivot4J

Question

I installed both the Saiku Analytics and the Pivot4J plugins on a Pentaho CE for a demo. From a high-level view, they both seem to be exactly the same, albeit Saiku seems to be performing a bit quicker. Can someone who is well acquainted with both please tell me the differences between the two and what factors I should take into consideration in selecting one over the other for Mondrain Cube analysis?

Answer 1

From my experience, Saiku seems a better experience, also it's supported by other plugins like IvyDD.

A deeper analysis was made by the developer of JPivot on this awnser:

Compared to Saiku, I think each project has its own advantage in different scenarios.

Saiku has a much lightweight architecture on the client side than our sample application and the plugin, so it can be deployed and embedded virtually anywhere.

While it's not much difficult to create a full REST style analytic application with Pivot4J, our current sample and plugin applications require at least a Servlet container to run and are more difficult to be embeded than Saiku in certain environment.

On the other hand, as Pivot4J is designed to be UI independent API from the start, it could provide more flexibility than Saiku in my opinion to developers when they want to build their own application on top of it, or intend to customize core behaviors of the API.

For example, if you want to use Pivot4J with your own application which is build with ExtJS, DhtmlX, or any other UI toolkits, it'd be much easier to achieve a seamless integration with Pivot4J, as it provides you with convenient abstract extension points to do that.

Finally, if you're familiar with Javascript you might find working with Saiku easier as it delegates most of the UI related works to the client side.

On the other hand, if you're an old school Java developer like me :) you might find our sample application to be easier to understand and work with, as there's virtually no custom script involved and everything is done on the server side with JSF component model.

To conclude, I'd like to say that Pivot4J is not just an API which cannot be used without writing much codes as it already includes quite feature complete Pentaho BI plugin for the upcoming 5.0 release of the platform. And as Pivot4J and Saiku take rather different approach from each other, each has its own strength and advantage which could be leveraged to suit the specific use case.

Source: https://stackoverflow.com/a/15264827/1557981