SO RELIABLE Iain Bray Sales Engineer InterSystems Corporation
What Is Knowledge Management? Increasing customer/partner/employee knowledge Rapid learning and redeployment of knowledge Increasing the value of intellectual property Adding unique values in products and services Creating new knowledge Sharing knowledge of work processes and innovations
The Problem All organisations have a vast store of useful information –In disparate systems Intelligent structures and common indices help but: –If it’s not a common question then the answer can take a while Indices use different names Data is stored in a different context No lack of information – just a lack of time to look at it!
What Do You Have To Do? Collaborate –Bringing together the disparate systems Search and Deliver –Find the right information and ‘tune-out’ the rest
Where Are We (Typically) DecisionSupportTransactionProcessing Simple Complex Caché Highest Performance, Most Scalable SQL for TP Complexity = Scalability x Type of Use
Collaboration (Joined Up Information) Relational Gateway –Projection of ODBC compliant databases as Caché Objects COM Gateway –Interface with COM applications through Caché Objects XML –Data and definition transfer with other applications EJB –Application Server Support Currently BEA Weblogic, JBoss, Borland Application Server IBM MQ Series –Support for message systems
Making Life Easier (Why Objects?) Relational Databases –Typically need two databases ‘Live’ database ‘Report’ database Two different database schemas More indices on the report database for better search speed Less indices on the live database for transaction speed Synchronisation problems Housekeeping and update problems Data not immediately available for reporting Adding object view is complex, costly and difficult to maintain
Making Life Easier (Why Objects?) Object Database –Only one schema –Online data instead of batch –Applications are ‘smarter’ because of availability of data –Business logic built into data structure instead of being ‘added on’
Joined Up Data Optimal storage mechanism Foundation for rich projections of the data Provides quick and easy access via industry standard APIs ActiveXODBC JavaWeb XMLEJB
What’s Coming Next Embracing OLAP as well as OLTP –Bit Map Indices Functionality in the product for a long time Now being integrated with SQL Performance improved –MDX (Multidimensional Expressions) Data cube creation and query
Bit Map Indices Extraordinarily quick boolean expressions –AND, OR, XOR etc. ‘Find all men who’s credit rating is ‘A’ and live in Slough’ sex=“M” AND credit=“A” AND postcode=“SL” Automatic creation and querying built into Caché 4.2
MDX – Multidimensional Expressions Microsoft –‘A highly functional expression syntax for querying multidimensional data ’ –‘OLAP Services supports MDX functions as a full language implementation for creating and querying cube data’ Functions for creating and querying cube data SELECT Measures.MEMBERS ON COLUMNS, [Store].MEMBERS ON ROWSFROM [Sales]
Data Cubes Caché data is already stored as sparse arrays Don’t necessarily need to create separate data cube Sparse arrays more efficient than data cubes –No redundant data
Making Life Simple Software developers understand their customers’ business models They need a database structure that will not frustrate the development of solutions for their customers Power and robustness through the sophisticated multidimensional data array Easy to exploit
Accelerating Innovation Always a strong database –Exploited and understood by professional developers Ever increasing number of access methods –Attracting new customers Increasingly innovative applications –Professional developers finding it easy to convert their ideas into practise
Summary Enabling the developer –Collaboration –Retrieval –Delivery Quick and Easy to develop and maintain –Quicker to market –Fewer development costs –More profit