The BIg Picture Dinsdag 17 september 2013
2 Agenda A short historical overview on BI Current Issues Current trends Future architecture First steps to this architecture
3 MIS/EIS Data Warehouse BI Multidimensional APL OLAP OLAP, MDX Database Report Files (R)DBMS Report Writer Historical view Easy to produce reports DWH Pivot Charts Slice and Dice, OLAP ETL concept Information presentation Layered Struct. Arch. Many loading options Dashboards, Maps Budgeting Traffic Lights Maps Analytics Drill down Slice and Dice Exceptions Events Backend Frontend Mainframe TSO Unix 3270 DOS PC Many reports Not flexible Data logistics Graphs Client Server Windows PC External data Analytics Integration budget/le Dashboards Database & Appl. server Web Performance Big Data Flexibility Too complex Data Sources Files Records ERP ERP+, Web 3rd Party
4 Current Architecture Device PC / Excel Web Mobile End user tool layer Reporting Budgeting Dashboards Data Layer DWH In-Memory Real-time virtual Data Sources ERP Other e.g. Excel 3rd Party
5 Proclarity Access Excel IS-AS-RS SQL Server Outlooksoft BusinessObjects KXEN SAP BW Sybase Crystal Reports Acta Tealeaf Informix TM/1 Adaytum Informatica Cognos Hadoop DataStage Lotus Google Focus DB/2 Ascential InfoSphere Netezza Roambi SPSS Qlikview SAS Oracle EveryAngle Oracle DWH Tableau Essbase Siebel Analytics MicroStrategy Sunopsis Spotfire Hyperion Brio Ingres
6 Current Issues BI Too rigid, inflexible Nice architecture, but takes too much time to change model No room for project data Time to report too slow Not all data is suited for a Data Warehouse (e.g. Big Data) Too complex to use Training required to empower users No BI Self Service, still dependent on IT Too Slow Users expect google like performance... Still not very Intelligent Past very well described, but how about the future? What information can be drawn from data?
7 Trends Consumerization Intuitive Mobile Easy and advanced Big Data 98% of today s information is digital Characterized by Some vs. All Data Accurate vs. Messy Causation vs. Correlation One road to the truth Data/information logistics with 3rd parties Uniform access to data Data concentration for all departments Google like performance In memory databases Map reduce & HDFS
Trends visualized 8
9 The era of 13-20 Desktop Web Mobile End user tool layer Reports Dashboards Visualization Analytical Applications Data Mining & Predictive Models Data Warehouse layer EDW (In Memory) Big Data DW Sandbox Data Lake Data Sources ERP CRM SRM Web Mail Sensors Robots Machine
10 BI BA Multidimensional OLAP, MDX In Memory network Database Report Layered Struct. Arch. Dashboards, Maps Traffic Lights In-Memory, fast! Data Mining, forecasting Sand boxes Big Data HANA, HIVE, HBASE Interactive Hybrid reports Analytics Exceptions Events Data Mining, Predictive Fcst, MapReduce Backend Database & Appl. server Cloud HDFS Frontend Web Tablet, Mobile Data Sources ERP+, Web 3rd Party MES, social media transactions
11 Use cases of Big Data Tracking outbreaks of seasonal flu based upon Google search archives (Google paper published in Nature in 2009) by comparing historical influenza data with the 50 million most used search terms between 2003 and 2008. First time right yield in manufacturing company: Joining large datasets from sensors and product data to understand current process capabilities for critical parameters and bring these parameters under control UPS: using sensors in cars to identify certain heat and vibrational patterns that in the past have been associated with failures in those parts Ad tracking: E-commerce sites record enormous amount of data which can, for instance, be used to track the effect of place, color, size, wording and other features of add s on websites. Mapping tweets to geographical maps and analyzing the emotions in those tweets to predict possible problem areas CAT Scan comparison to facilitate the automatic diagnosis of medical issues
SAP filling in the gaps 12
13 Each company roadmap will be different What would be a case for predictive What would be Useful on mobile End user tool layer Do I need all this data on premise? What can enable self service? Data Warehouse layer What data would be handy In Memory? Data Lake What data would be handy in Real Time? Sources What data would be smart to analyze but is to large?
Start taking small steps: Create an one KPI dashboard on mobile using BI on demand 14
Start taking small steps: Provide that self service BI insight with Lumira on a Hana data set in the cloud 15
Start taking small steps: Gather big data insight with your Google or Amazon big date cloud environment 16
Start taking small steps: Retrieve real time data into a HANA environment in the cloud and start analyzing, monitoring and reporting 17
Finally: It is about catching the right technology wave, mobile, cloud, big data, self service BI, but that only works if you try 18