210 likes | 304 Views
Introduction to OLAP and Analysis Services from Microsoft (ONLY for INTERNAL USE). Josef Schiefer IBM Watson Research Center josef.schiefer@us.ibm.com. What is OLAP?. Online Analytical Processing - coined by EF Codd in 1994 paper contracted by Arbor Software *
E N D
Introduction to OLAP and Analysis Services from Microsoft(ONLY for INTERNAL USE) Josef Schiefer IBM Watson Research Center josef.schiefer@us.ibm.com
What is OLAP? • Online Analytical Processing - coined by EF Codd in 1994 paper contracted by Arbor Software* • Generally synonymous with earlier terms such as Decisions Support, Business Intelligence, Executive Information System • OLAP = Multidimensional Database • MOLAP: Multidimensional OLAP (Arbor Essbase, Oracle Express) • ROLAP: Relational OLAP (Informix MetaCube, Microstrategy DSS Agent)
OLAP is FASMI • Fast • Analysis • Shared • Multidimensional • Information Nigel Pendse, Richard Creath - The OLAP Report
Cubes • A cube stores information in a multidimensional structure and is the central object in a multidimensional database. • Each cube contains a set of dimensions and measures. • Dimensions are derived from the tables and columns in your data that provide the categories you want to analyze. • Measures are the quantitative data derived from your data columns
Dimensions • The dimensions you build should be distinct categories you want to add to cubes in your OLAP database. • example: geography, time, or employee dimensions represented in the picture
Product Region Time Industry Country Year Category Region Quarter Product City Month Week Office Day Region W S Juice N Cola Product Milk Cream Gel Soap 1 2 3 4 5 6 7 Month Cube and Dimensions Dimensions: Product, Region, Time Hierarchical summarization paths
Dimensions and Hierarchy • Dimensions are the categories used to organize or describe analysis information • Dimensions are used to navigate the information and to summarize the details into more aggregate data. • Frequently used dimensions include time periods, geography, products, organization, and so on. • Often dimensions are hierarchical (World - Continents - Countries)
Measures =numercial Values • Measures are the quantitative data in an OLAP database. • For example, values such as sales, budget, cost, and so on, are all examples of measures. • Measure values are organized in data cubes according to dimensions
Aggregations • Aggregations greatly improve query efficiency and response time. A cube can hold a number of aggregations. • The aggregation amount is based on several factors - the size of the data, the amount of storage space you allocate for aggregation storage, the mode of storage you select, and how much you want to optimize the aggregations design.
Primary OLAP Problems • Rigid, inflexible architectures • MOLAP or ROLAP • Significant scalability problems • Data explosion and sparsity • Poor distributed client/server implementation • Separation of data warehousing from OLAP tools • Lack of integration between user tools and OLAP • Difficult to prototype, develop, deploy • Time and expense
MS-AS: Architecture Microsoft Analysis Services are optimized for all OLAP architectures and offers seamless integration • MOLAP: aggregations & details managed in an efficient multidimensional store • ROLAP: aggregations created in relational store • HOLAP: different things to different vendors • Aggregations: details in relational, aggregations in MOLAP store • Partitions: single logical cube physically divided into multiple MOLAP and ROLAP partitions • Virtual cubes: “view-like” join of multiple MOLAP and ROLAP cubes
MS-AS: Scalability • MS-AS offer major innovation • Data explosion managed by partial pre-aggregation • Automatic elimination of sparse storage • Partitioned cubes • parallel query processing across clustered servers • fine tuning of aggregations, to better manage performance and disk space trade-offs
MS-AS: Scalability • Cooperative client/server query management and caching • network traffic minimized • server queries processed efficiently • Microsoft Data Cube Service • desktop component ships with next release of Office • used with Excel, Access, and Web • supports local, offline usage
Microsoft Data Cube Service • Basic architecture: • Cache query results and metadata, not disk pages. • Algorithms deduce missing data and transform queries • Aggregation • Filtering • Combination • Instant reply to cached queries
MS Data Cube Benefits • Efficient distribution of query and calculation processing across client & server • Single component spans Microsoft desktop and server platforms & products • Unifies the MD data access story across Excel, MS-AS, and SQL Server • Enables Microsoft to establish industry standard for MD data access • Basis for MS-AS and Excel mobile story
MS-AS: Integration • The Microsoft Analysis Services integrate the maintenance of OLAP with the underlying data warehouse • Design the DW structure • Create the DW tables/cubes • Populate the DW tables/cubes • Maintain by incremental loads • Optimize by actual usage patterns • Manage users, scripts, usage, metadata • Multiple data sources (not just SQLS)
MS-AS: Integration • OLE DB for OLAP & ADO MD • based upon existing data access technology • establishes industry standard for MD data access • OLE DB/ODBC enable MS-AS to access data in all major RDBMs • Third party client applications
OLAP Problem: Complexity • OLAP products are traditionally difficult to configure, develop, and deploy • Arcane tools • Heavy consulting • Poor integration
Client Tier ADO MD • Data selection & navigation • Presentation and charting • What-if formulas • Client side caching • Desktop object model • Offline usage • Excel • ActiveX Controls • Third Party Applications OLE DB for OLAP DCube MS-AS Server • Multidimensional calcs • MOLAP/ROLAP/HOLAP data Modeling/aggregations • Security • Metadata management • Server side caching • Administrative tools • Server object model • Query distribution Data Warehouse Tier MOLAP ROLAP • MS-AS Server • SQL Server HOLAP DTS OLE DB OLTP Source Tier • RDBMs 3 Tier Architecture & Components