Business Intelligence. Cubed.

Develop and Deploy Analytics. Free End-Users and IT. The way it should be.

The SiSense PrismCubed business intelligence platform provides all the functionality and feature-set required to rapidly develop and deploy scalable business intelligence applications. Its core components are the PrismCubed application modeling environment and the ElastiCube engine. The PrismCubed application modeling environment is where applications are designed and developed. It is a highly visual product where the application is designed by dragging and dropping widgets onto a canvas, binding them to data and defining application logic through an intuitive set of tools.

Unified Analytic Models

PrismCubed internally utilizes SiSense technology called Unified Analytic Models. These models enable the representation and querying of data in dimensional form, even if that data is not structured dimensionally. Dimensional models are often encountered in OLAP based solutions and are typically queried using multidimensional expressions (MDX). PrismCubed's Unified Analytic Models can be used to query OLAP cubes, but they can also be (and usually are) used over SQL-based systems.

Unified Analytic Models can be completely translated to a single standard SQL query, although they sometime are not for performance considerations. This fact ultimately means that PrismCubed applications can execute queries directly over any SQL-based database.

The above amount to the fact that the application modeler need not care whether the application data resides in an SQL Server database, an Oracle data warehouse or an ElastiCube. This also means that the modeler does not need to know SQL or any other query language to formulate queries. The queries are visually defined over the dimensional model and the required SQL queries are generated internally.

ElastiCubes

PrismCubed is powered by SiSense technology called ElastiCubes that enables it to process huge amounts of data (practically unlimited under a 64-bit environment) without complicating the solution's architecture or increasing hardware requirements.

From the user's perspective, ElastiCubes are a regular SQL-based relational data source just like any other. However, diving into the internals of an ElastiCube would reveal that the underlying mechanics are completely different. These inherent differences are what make ElastiCubes perform so well, up to 50-100 times faster than standard relational databases, under the same hardware configuration.

ElastiCubes use a combination of column-based storage and in-memory query processing to return query results in seconds, even when dealing with huge amounts of data. For storage and persistency ElastiCubes store data as individual fields, rather than row-by-row as relational databases do. These fields are stored on the hard-drive and quickly loaded on demand when particular queries require them, after which they are disposed freeing up the memory used and making it available for other queries.

When an ElastiCube receives a query to execute, fields are loaded into memory as the query requires them and disposed of when they are no longer needed. Fields not explicitly referenced by a particular query will not take up any memory. Typical business intelligence queries often use only a subset of all available fields, therefore this mechanism usually leaves enough RAM to perform complex calculations over large amounts of data entirely in-memory, without intermediate hard-drive access to slow it down.

Unlike OLAP technology, ElastiCubes do not require to pre-calculate or pre-aggregate anything in order to return results quickly. This is why ElastiCubes take a fraction of the time to create and process than their OLAP based counterparts.

Unlike in-memory database technology, ElastiCubes do not load the entire data into memory. This fact makes an ElastiCube's size to be practically unlimited. ElastiCubes can have as many fields as required, and this would have no impact on overall performance due to the fact that fields are only loaded into memory when a particular query needs them. In addition, this architecture makes ElastiCubes a viable solution on a simple desktop, not only a server machine.