[ PREVIOUS ARTICLE | Table of Contents | NEXT ARTICLE ]

DATA MINING WITH THE EXPLORATION WAREHOUSE: PART III
by W H Inmon


There are then many obstacles awaiting the data miner. Having an exploration warehouse as a standard part of the DSS infrastructure allows the data miner to turn the data upside down looking for these important relationships.

The exploration warehouse provides the data miner with the infrastructure needed for success.

THE SPEED OF QUERIES

One of the important beneficial aspects of the exploration warehouse is the speed of querying afforded by having token based "in memory" data. Response times to the data miners query that can be achieved are simply breathtaking. While these speeds of query are technologically impressive, they have a serious and very positive business consequence. The speed of analytical processing that the exploration warehouse affords the data miner is something that greatly facilitates the analytical process.

Consider a data miner that can only do one query a week. The data miner must carefully craft his/her query because it will be a long time until the query can be reprocessed and the results can be obtained. But what if the data miner can get two to three second response time? In this case the data miner has the room to experiment. The data miner can be creative and can explore hunches. If a hunch does not pan out, then only a few minutes have been wasted. The speed of query processing made possible by an exploration warehouse allows the data miner to exercise creativity and intuition in the analytical process. And it is here that a data miner is at his/her best. There is then a real importance to the very fast speed of processing that a data miner can do in an exploration warehouse environment.

Sidebar to the exploration warehouse -

So what vendors are active in the exploration warehouse and token based technology arena? There are several vendor implementations of token based technology and exploration warehouses that are in states of anywhere from in progress laboratory development to market ready. Some of the leading vendors are -

Sand Technology/Hitachi Data Systems, Montreal Canada. Sand Technology's patented Nucleus Exploration Series is a full blown end to end exploration warehouse/mart token based implementation. Nucleus Exploration Warehouse and Mart provides powerful analytical processing for iterative, ad hoc, and forensic queries available on both 64 bit Unix and Intel based 32 bit NT operating environments.

HOPS, Inc, Miami Lakes, FL. HOPS supplies the foundation for exploration warehouses. HOPS compresses data and executes highly optimized queries. When asked whether token based compression was used, a company spokesman declined to comment. The company spokesman described HOPS compression techniques as a data base pattern compression with run encoding. HOPS, like Nucleus, is a commercially ready product.

An unidentified source for Compaq Corporation, Sugarland, Tx. says that token based technology has been developed in the German laboratories of Tandem computers (now a Compaq company). Compaq's token based technology is said to still be in a developmental mode and it is not know when and if Compaq will introduce their version of token based technology in support of exploration warehouses to the marketplace.

A last foundation for exploration warehouses is Sybase's IQ product. Sybase IQ shares some very similar features with token based technology but technicians state there are some fundamental differences between token based technology and Sybase IQ. Sybase IQ supports bit map processing but not the domain bit indexes that are found in a token data base. In addition technicians state that Sybase IQ forces the designer to choose index types at the moment of design based on the grouping of data, cardinality of the contents of the index or compression, depending on the data. In addition, Sybase IQ operates in a write once mode, not a fully operational update mode. Having stated some of the differences between Sybase IQ and true token based technology, Sybase IQ is a platform worthy of consideration for the exploration warehouse.

---

For more information, see http://www.pine-cone.com


[ PREVIOUS ARTICLE | Table of Contents | NEXT ARTICLE ]