|
OLAP Load balancing
Load balancing is a method by which an application OR server ensures that all resources are equally loaded so to deliver the best performance. This load balancing is achieved by balancing the load across the processors, across the servers (in a cluster), across the threads and across the partitions.
OLAP Usage based optimization
An OLAP tool should be able to read through the query log and should be able to optimize the aggregations and OLAP vs. Data Warehouse cut, on the basis of the optimization parameters. Various optimization parameters can be:
- Queries for certain set of users.
- Queries which take more than a certain length of time.
- Queries which are most used.
Multi-threading
Multi-threading is a known technology concept, whereby an application spawns multiple threads to parallelize the work. For example, instead of having one process searching a table of 1 million rows, it can spawn ten threads with each parallel searching the chunks of one hundred thousand records.
Parallel processing
This is different from multi-threading, as this is typically managed at a database level OR at a server level, where the parallel processing is done on multiple database partitions OR across processors.
Ability to perform OLAP calculations at run-time
This is important because the requirements of ad hoc queries are often unanticipated by designers and may not have been directly included in the aggregation OR compilation schemes of OLAP servers. The ability of a server to perform calculations rapidly at run-time is a critical determinant of ad hoc query capability.
OLAP Performance on Ad-hoc queries
OLAP solutions face bigger challenges with front-end analytic applications posing unpredictable and large ad-hoc queries. The OLAP solution design should allow an acceptable query performance through the efficient database design; query optimizing, multi-threading and other methods.
As you evaluate the OLAP server tools, you should ask for the performance bench-marks from the Vendors.
|
