File Library

Essbase technology is around for more than two decades and compute power has significantly changed, moving from 16 bits to 64 bits and also supporting multiple cores and Terabytes of memory. Also, some of the fundamental assumptions of OLAP that existed when it was invented are no longer valid: small number of dimensions, dimensions with small cardinality, you have entire weekend to build the cube, OLAP is all about pre-computing aggregates, data refresh happens once a week, meta data refresh doesn’t happen often, reporting and planning systems could be different and one can export from planning to reporting cubes and wait times are acceptable, and so on.

Now we have requirements to support 7x24 world with no downtime, very fast data refresh, large number of dimensions and with dimension being large, real time reporting against planning systems, fast metadata refresh, and so on. All these requirements call for OLAP technology that doesn’t do any pre-computation of aggregate and calculated measures. With data refresh all the pre-computations are invalid anyway. We need to dynamically compute everything and this calls lot of innovations cube data structures for metadata, calculation algorithm and aggregation algorithm. Also, this demands all the metadata and operands being in memory.

We will highlight architectural innovations in Essbase that addresses all these challenges.