We have been working toward in-memory analysis capabilities, within JasperServer, for a couple of years. With our previous major release, Jaspersoft 3.5, we made an important declaration: in-memory analysis techniques are going to be a very important part of our future product direction. Being an open source BI company means a commitment to innovative technology and value for customers. An in-memory strategy gives customers, both enterprises and ISVs, more options to improve their decision making. So, you’d expect that Jaspersoft 3.7 would meaningfully advance our Server’s in-memory, multi-dimensional analysis capabilities. What you might not expect is just how far it goes.
Firstly, in-memory analysis enables JasperServer to hold the query result data set in memory, complete with sufficient meta-data that enables the user to continue analyzing the data along multi-dimensions, without having to re-issue a new query. And, the user manipulates the results using an enhanced data explorer interface, which is a simple, purely web-based, drag-and-drop environment. So, the outcome is a fast and friendly way to sort, filter, pivot, compare, and drill into data without having to separately design and maintain OLAP cubes or having to use a special front-end tool to interact with those special data sources.
What can be done with Jaspersoft 3.7’s new in-memory analysis? The screen shots below provide an example of a cross-tab report, complete with multiple dimensions being viewed (Locations in the form of States and Cities and Product by Name and Type) and a pie chart that graphs some of the data from this report (sales summary by State). The user can add dimensions and measures just by dragging and dropping and then re-orient or adjust existing dimensions by “pivoting” . This cross-tab report can include summary calculations, calculated fields, sorting and filtering. And, the user can continue clicking on data until he or she reaches the ultimate transaction, should they wish.
So what does In-Memory Analysis mean for Jaspersoft customers”? Unlike other traditional and in-memory BI vendors, we’re taking the best of both worlds approach whereby the system administrator can configure the server to perform processing either against the in-memory data set, or alternatively, push processing down to the underlying data store. The decision on which approach is optimal for a given deployment will depend a lot on the query performance characteristics of the data store. For example, a traditional OLTP or non-relational data store may benefit significantly from in-memory processing, whereas a query optimized analytic data store may provide performance similar to or better than in-memory processing. Combining this flexible architecture with the cost advantages of not using an OLAP server gives customers choice and a BI platform that can grow as their data and analysis requirements do.
This is just a summary of the many in-memory analysis capabilities we’ve included in Jaspersoft 3.7. I invite you to check out some additional information and a demonstration here.
Next post, I’ll describe our quest to provide rich, Flash-based visualization to make every report, dashboard, and web application come to life.
Chief Executive Officer
2 years ago