Download presentation
Presentation is loading. Please wait.
1
A New Storage Engine Specialized for MOLAP
MySQL Users Conference 2006 Dr. Juergen Faisst Thinking Networks AG Thinking Networks AG
2
Thinking Networks AG Specialist for high-end planning, budgeting and consolidation solutions Software: TN Planning, > users Based in Aachen, Germany 50 employees 7 Mio$ revenues Thinking Networks AG
3
A New Storage Engine Spezialized for MOLAP
World’s Largest Tourism Group Revenue Management Simultaneous planning of products, destinations, flights Forecast based on Neural Networks Financial Planning of Cost Human Resources Investments A New Storage Engine Spezialized for MOLAP Thinking Networks AG
4
A New Storage Engine Spezialized for MOLAP
Session Structure Motivation Why do Planning and Budgeting Applications need a different kind of storage engine? The THINKING NETWORKS Storage Engine What makes our storage engine different? Project Status Where are we now? Where do we want to go? A New Storage Engine Spezialized for MOLAP Thinking Networks AG
5
Motivation Why do Planning and Budgeting Applications need a different kind of storage engine?
6
Business/Corporate/Enterprise Performance Management at a glance
Motivation Business/Corporate/Enterprise Performance Management at a glance Performance Management Analysis incl. Data Mining Stat. Consol. ROLAP: Mondrian OLAP4All HydraCube with embedded BerkleyDB Similar to Business Objects Cognos MicroStrategy Reporting Planning & Budgeting Data Warehouse ROLAP MOLAP-DBMS RDBMS A New Storage Engine Spezialized for MOLAP Thinking Networks AG
7
MOLAP storage speeds up Reporting and Analysis
Motivation MOLAP storage speeds up Reporting and Analysis Hybrid engines of large database vendors focus on performance and scalability while saving disk space MS SQL Server Analysis Services Oracle10g OLAP Option ( ) IBM DB2 OLAP Server See for more details A New Storage Engine Spezialized for MOLAP
8
MOLAP storage empowers Planning & Budgeting
Motivation MOLAP storage empowers Planning & Budgeting Specialized database vendors focus on MOLAP calculation engine for aggregation, allocation and business rules TM1 Essbase Alea Express TN Planning A New Storage Engine Spezialized for MOLAP
9
A New Storage Engine Spezialized for MOLAP
Motivation A simple example for the need of a MOLAP calculation engine: Sales planning Measures: Sales Volume, Price, Revenue Scenario: Plan, Actual Time: Months < Quarters < Year Product: Products < Product Groups < Total Location: Store < State < Country < Total A New Storage Engine Spezialized for MOLAP
10
A New Storage Engine Spezialized for MOLAP
Motivation Reporting is easy A New Storage Engine Spezialized for MOLAP
11
A New Storage Engine Spezialized for MOLAP
Motivation Planning is different A New Storage Engine Spezialized for MOLAP
12
Modeling the business logic in a planning application is a challenge
Motivation Modeling the business logic in a planning application is a challenge Custom aggregation rules, e.g. The ”price-volume-amount”-problem The ”value type”-problem (Legal) consolidation Custom allocation rules, e.g. Planning on aggregate levels The ”allocate with exceptions”-problem Custom calculation rules, e.g. Currency conversion A New Storage Engine Spezialized for MOLAP
13
A New Storage Engine Spezialized for MOLAP
Motivation What about MySQL ? “ MySQL’s pluggable storage engine architecture offers the unique opportunity to build a hybrid Data Warehouse using a powerful MOLAP storage and calculation engine for Planning & Budgeting” - THINKING NETWORKS A New Storage Engine Spezialized for MOLAP
14
The THINKING NETWORKS Storage Engine
What makes our storage engine different?
15
Relational addressing of data: Searching for the right record
THINKING NETWORKS Relational addressing of data: Searching for the right record Keys and indexes A New Storage Engine Spezialized for MOLAP
16
A New Storage Engine Spezialized for MOLAP
THINKING NETWORKS Multidimensional addressing of data: Direct access in a coordinate system Location Product California Printer Time Feb A New Storage Engine Spezialized for MOLAP
17
Major advantage and disadvantage of multidimensional addressing
THINKING NETWORKS Major advantage and disadvantage of multidimensional addressing Very fast access, independent from data volume Very fast access, independent from data volume is needed for a powerful calculation engine Data explosion in sparsely filled data spaces A New Storage Engine Spezialized for MOLAP Thinking Networks AG
18
Avoiding data explosion by preventing from storing empty cells
THINKING NETWORKS Avoiding data explosion by preventing from storing empty cells Sparse Sparse Sparse Dense A New Storage Engine Spezialized for MOLAP
19
A New Storage Engine Spezialized for MOLAP
THINKING NETWORKS Sparse rows in a reference table with a fast hash algorithm pointer to … … high density matrixes Only matrixes containing at least one non empty cell are stored physically A New Storage Engine Spezialized for MOLAP Thinking Networks AG
20
A New Storage Engine Spezialized for MOLAP
THINKING NETWORKS A cell, a row, a column or the whole matrix can be read or written by one single access Only matrixes containing at least one non empty cell are stored physically A New Storage Engine Spezialized for MOLAP Thinking Networks AG
21
Project Status Where are we now? Where do we want to go?
22
SQL-read-access to THINKING NETWORKS Storage Engine via MySQL
Project Status SQL-read-access to THINKING NETWORKS Storage Engine via MySQL Thinking Networks A New Storage Engine Spezialized for MOLAP
23
Multi-dimensional read and write-back access via TN Planning
Project Status Multi-dimensional read and write-back access via TN Planning TN Planning Custom Tailor Navigator Language Processor DML, DDL, Views, Rules, Stored Procedures, Storage Engine Thinking Networks A New Storage Engine Spezialized for MOLAP
24
A New Storage Engine Spezialized for MOLAP
Project Status MySQL AB considers support of multi-dimensional queries and handler calls SELECT ( ( [scenario].[Actual],[time].[Jan]), ([scenario].[Plan],[time].[Feb] ) AS COLUMNS ( [items].[Sales], [items].[Price], [items].[Revenue] ) AS ROWS FROM SalesCube WHERE ( [location].[California], [product].[printer]) MySQL (MD)SQL Interface, Parser, Optimizer MDX standard defined by Microsoft, Hyperion and others SQL Analytic Extensions is Oracle‘s way to open SQL for multidimensional queries int get_cube_slice ( char slicespec, int nof_rows, char **rowspec, int nof_cols, char **colspec, double *results ) A New Storage Engine Spezialized for MOLAP Thinking Networks AG
25
We would like to encourage the developers of other storage engines ...
Project Status We would like to encourage the developers of other storage engines ... … to support the new multi-dimensional handler calls MDX Thinking Networks A New Storage Engine Spezialized for MOLAP
26
A New Storage Engine Spezialized for MOLAP
Target architecture TN Planning Custom Tailor Navigator Language Processor MDX Thinking Networks A New Storage Engine Spezialized for MOLAP
27
Thinking Networks considers to offer …
Project Status Thinking Networks considers to offer … THINKING NETWORKS Storage Engine for MySQL under GPL TN Planning as commercial software Language Processor Custom Tailor Navigator Support & Maintenance Package Thinking Networks A New Storage Engine Spezialized for MOLAP
28
Let MySQL become the leading database
for Business Intelligence and Analytic Applications Thank You for Your Attention! Dr. Juergen Faisst Thinking Networks AG / A New Storage Engine Spezialized for MOLAP
Similar presentations
© 2025 SlidePlayer.com. Inc.
All rights reserved.