|
This pages talks about the scalability from user point of view. The scalability and performance is not only dependent upon the OLAP server tool, but also on the way it is designed and configured.
Size of the database in OLAP server:
The basic design of the database of the server limits the data it can store. For example MOLAP servers may not be able to store the data in the range of terabytes.
Processing the cubes out of the Data Warehouse:
An OLAP server has the programs which pick the data from the Data Warehouse and process it into cubes. The efficiency of these programs drive on how fast the cubes can be refreshed (the size of the hardware being a constant).
Ability to handle large number of dimensions in a Cube
This is to do with ability to add as many dimensions as required in a single cube. This is again where a ROLAP/HOLAP scores over MOLAP. MOLAP servers are limited by a set of dimensions in the cubes, with-in which it can work efficiently.
Ability to handle large size Cube dimensions
This is in terms of the number of members as dimensions. Here again ROLAP and HOLAP score over MOLAP. A MOLAP OR MOLAP component of a HOLAP (HOLAP is a hybrid of MOLAP and ROLAP). A ROLAP OR HOLAP will have the lowest level instances of a dimension (For example each sales lead in the sales lead dimension), which means that a dimension can have millions of members at the lowest level. An OLAP solution should be able to handle the exponential growth of the size of large dimension, by effective storage management.
Ability to handle large number of users
An OLAP should be able to handle large number of users, without a major impact on the response time.
Ability to handle large, varying and frequent queries
The number, type and complexity of queries vary from time to time. OLAP systems are severely overloaded around a period end, and also during the planning and forecasting cycles. This may not be desirable, as you can leverage BI much more than period-end reporting/performance score-carding OR business planning. Irrespective, an OLAP solution should be able to handle frequent load-peaks and also ad-hoc and unpredicted queries. |
