Consumerization of BI - Bring Your Own Insight

Editor's Notes

1. Maar zodra je zo’n document weghaaltkomterweereennieuwevoor in de plaats of zelfsmeer, lijktweleenbeetje op het verhaal van Hydra (1 kop afhakkenkomenermeerderevoor in de plaats)
2. Userszeggentegen IT dat het snellermoet : het had gisterenafgemoeten!
3. ONZE implementatie van CEPBEST BEWAARDE GEHEIM in SQL SERVERSTREAMING DATA ANALYSERENCLOUD VERSIE is in de MAAK
4. Steep Learning Curve; Slow & InefficientToday, if you want to do insights with Big Data, you would need to do one of two things. The first is to be able to have a mechanism to query HDFS. This is done around learning the ecosystem of Hadoop like HDFS, MapReduce, Hive, Hbase, etc. While this is not a bad thing to learn, most business users will want to use their standard BI query tools to get access to the data. They don’t want to learn map reduce or use a separate BI tool to get access at the data. The other mechanism is to have IT manually move HDFS into the data warehouse and join the data at the warehouse level. This mechanism will require this additional step of IT understanding the requirements of the business and building this step prior to any analysis that can be done by the business.
5. Microsoft solves this limitation through a new feature in SQL Server 2012 PDW called PolyBase. For PolyBase, we significantly improve how traditional data warehouses interacts with Hadoop. It enables integrated query across Hadoop and relational data. Without a prior manual intervention (moving the data by IT), PolyBase Query Processor can accept a standard SQL query and join tables from a relational source with tables from a Hadoop source to return a combined result seamlessly to the user. There are two main benefits around this:The user only issues a standard SQL Query. They don’t need to learn MapReduce. This means, standard BI tools can be used to query structured and unstructured data togetherExample Standard SQL query: select c from hdfsCustomer, o from PDW where c.c_custkey = o.o_custkeyIT doesn’t need to do a prior mapping and moving of data from HDFS into the warehouse.  PolyBase was pioneered in Created in Jim Gray Systems Labs by David DeWitt, former Professor Emeritus of Computer Sciences. Dr. DeWitt is known for pioneering research in parallel databases, benchmarking, object-oriented and XML databases. He has over 120 papers published. Details of how this work is as follows:When the user uses PolyBase, they can see the tables in Hadoop through external tables. They can then select the tables they want in Hadoop and join them with PDWPolyBase will then import/export Hadoop data to and from PDW in parallel for performance (using DMS on each PDW compute node) Unstructured data that returns will be temporarily or permanently stored on PDW and compute node does all processing to join and return back results Additionally, Microsoft also released a Windows-based Hadoop distribution (HD Inisghts) both on premise on Windows Server and on the cloud with Azure.   
6. Start Small With A Few TB and Linearly Scale OUTSQL Server 2012 PDW can scale from the smallest to the largest data warehouse requirements that you have. This means if you have a capacity requirements anywhere from 0TB up to 5 Petabytes, SQL Server 2012 PDW has got you covered. As mentioned before, this means you can incrementally add capacity to your existing implementation to rapidly scale out to the largest deployment. However, it also means you can start small with only a few Terabytes.
7. Better word for big data would be ambient data.However, it is not about data it’s about the value that you get from it. The companies that will be able to faster get better value out of ambient data will be the most successful.
8. Sums up Microsoft’s BI vision:Our vision is not only about Big Data, but also around Small Data and Any Data, regardless of what type it is and where it is stored.----Accessible anywhere: on any device at any time----Yesterday, today, tomorrow: not just looking back, but also look at what is happening now and predict the future--This is apparent in multiple sides of our products: functionality, usability, speed, visualization, etc…