FAST has entered a market that is becoming overcrowded and is already highly competitive. It is particularly so in the data warehousing market where there has been a stream of new and innovative solutions from the likes of Netezza and Datallegro. To add to those, HP recently entered the market with Neoview, a dedicated enterprise data warehouse that is aimed at the high end of the enterprise market dominated by Teradata and IBM.
AIW offers a totally new take on an old problem, and as such could prove to be disruptive. It combines the power of FAST’s search technology, its linguistic-based data quality capabilities, and Radar, its acquired ROLAP solution.
Key components of the FAST AIW platform include FAST Radar and the FAST Data Cleansing Solution.
FAST Radar is a ROLAP BI tool that provides multidimensional aggregation and analysis, using a patented process called Pyramid that speeds up aggregation for analysis by grouping only the dimensions that are necessary at each level of the Pyramid. The Pyramid Server has alert and scheduling agents that perpetually monitor all user-defined thresholds and scheduled events, and broadcast requested information when the user-defined criteria is reached. The product is browser-based with a zero footprint on the client side.
The FAST Data Cleansing Solution provides a search-based approach to collecting and cleansing entities and unique identifiers from data repositories in any format and from any location, to create a single master index. It uses techniques such as structured data searches with contextual awareness, fuzzy and phonetic matching, and rules-based cleansing.
Linguistic data cleansing techniques work well for search purposes, but they are as yet to be proven for delivering accurate enough information for decision making. However, if BI is to include unstructured data, then that data will require checks and cleansing too. Traditional data quality techniques do not apply to unstructured data, therefore a new linguistic-based approach may well be the answer.
The AIW is an interesting proposition. It could prove to be faster to deploy than the traditional data warehouse and potentially easier to change post-deployment. Also, with a data warehouse you have to know what queries you are likely to generate later so that the right data can be pulled into the warehouse. The search index can potentially overcome this issue too, proofs of concept will be eagerly awaited. In the meantime, the new paradigm is likely to spark more developments in the BI sector, leading to additional uses of the search index.
Content from our partners
Source: OpinionWire by Butler Group (www.butlergroup.com)
