SYBASE IQ PROVIDES ADVANCED ANALYTICS TO UNLOCK CRITICAL BUSINESS INSIGHTSDIAL NUMBERS:1‐866‐803‐21431‐210‐795‐1098PASSCOD...
YOUR HOSTS FOR TODAY             Your Host…                            Guest Speaker…                                David...
HOUSEKEEPING        Questions?        Submit via the ‘Questions’ tool on your Live Meeting console, or           call 1‐86...
Big data and the warehouse        Philip HowardResearch Director – Bloor Research                            …optimise you...
Agenda                        What is different about big data?                        Why would you want to take a big da...
How is big data different?Confidential © © Bloor Research 2012 Confidential Bloor Research 2009                    telling...
But …Confidential © © Bloor Research 2012 Confidential Bloor Research 2009         telling the Information Management stor...
But …Confidential © © Bloor Research 2012 Confidential Bloor Research 2009         telling the Information Management stor...
But …Confidential © © Bloor Research 2012 Confidential Bloor Research 2009         telling the Information Management stor...
2 types of big dataConfidential © © Bloor Research 2012 Confidential Bloor Research 2009                 telling the Infor...
Structured dataConfidential © © Bloor Research 2012 Confidential Bloor Research 2009               telling the Information...
Unstructured dataConfidential © © Bloor Research 2012 Confidential Bloor Research 2009                telling the Informat...
Why - unstructuredConfidential © © Bloor Research 2012 Confidential Bloor Research 2009                telling the Informa...
Why - structuredConfidential © © Bloor Research 2012 Confidential Bloor Research 2009               telling the Informatio...
How - Hadoop & NoSQL                                              Key-value model:                                        ...
NoSQL vs Triple StoresConfidential © © Bloor Research 2012 Confidential Bloor Research 2009                  telling the I...
How – Hadoop basicsConfidential © © Bloor Research 2012 Confidential Bloor Research 2009                 telling the Infor...
MapReduceConfidential © © Bloor Research 2012 Confidential Bloor Research 2009           telling the Information Managemen...
But …Confidential © © Bloor Research 2012 Confidential Bloor Research 2009         telling the Information Management stor...
Alternatives/extensions to                                                 HadoopConfidential © © Bloor Research 2012 Conf...
Hadoop and the warehouse                         Technology                                             Results and Applic...
Hadoop and Sybase IQ                                  For example, in telecoms you might use a data warehouse to          ...
Hadoop and the warehouse                                                                                Results and Applic...
Hadoop and Sybase IQ                                       An example use case is in eCommerce, where clickstream         ...
Hadoop and the warehouse                       Technology                                                      Results and...
Hadoop and Sybase IQ                                                    POS Data                 Inventory Data           ...
Hadoop and the warehouse                         Technology                                                 Results and Ap...
Hadoop and Sybase IQ                                                  Smart Grid                        Smart Meter       ...
Hadoop and Sybase IQ                     NoSQL scale out is easier and cheaper                     than scale up          ...
Conclusion                We can now analyse data in ways that were not                previously possible that can provid...
THANK YOU         FOR MORE INFORMATION         WWW.SYBASE.COM/SYBASEIQBIGDATA 31 – Company Confidential – March 27, 2012
Sybase IQ Advanced Analytics with Big Data
Upcoming SlideShare
Loading in …5
×

Sybase IQ Advanced Analytics with Big Data

1,301 views

Published on

Published in: Technology, Business
0 Comments
1 Like
Statistics
Notes
  • Be the first to comment

No Downloads
Views
Total views
1,301
On SlideShare
0
From Embeds
0
Number of Embeds
3
Actions
Shares
0
Downloads
0
Comments
0
Likes
1
Embeds 0
No embeds

No notes for slide

Sybase IQ Advanced Analytics with Big Data

  1. 1. SYBASE IQ PROVIDES ADVANCED ANALYTICS TO UNLOCK CRITICAL BUSINESS INSIGHTSDIAL NUMBERS:1‐866‐803‐21431‐210‐795‐1098PASSCODE: SYBASE
  2. 2. YOUR HOSTS FOR TODAY Your Host… Guest Speaker… David Jonker Philip Howard Product Marketing Research Director,  Sybase, an SAP Company Bloor Research2 – Company Confidential – March 27, 2012
  3. 3. HOUSEKEEPING Questions? Submit via the ‘Questions’ tool on your Live Meeting console, or  call 1‐866‐803‐2143 United States, 1‐210‐795‐1098 Other  Password SYBASE  Press *1 during the Q&A segment Presentation copies? Select the printer icon on the Live Meeting console3 – Company Confidential – March 27, 2012
  4. 4. Big data and the warehouse Philip HowardResearch Director – Bloor Research …optimise your IT investments
  5. 5. Agenda What is different about big data? Why would you want to take a big data-based approach? How would you deploy it using Sybase IQ?Confidential © © Bloor Research 2012 Confidential Bloor Research 2009 telling the Information Management story telling the right
  6. 6. How is big data different?Confidential © © Bloor Research 2012 Confidential Bloor Research 2009 telling the Information Management story telling the right
  7. 7. But …Confidential © © Bloor Research 2012 Confidential Bloor Research 2009 telling the Information Management story telling the right
  8. 8. But …Confidential © © Bloor Research 2012 Confidential Bloor Research 2009 telling the Information Management story telling the right
  9. 9. But …Confidential © © Bloor Research 2012 Confidential Bloor Research 2009 telling the Information Management story telling the right
  10. 10. 2 types of big dataConfidential © © Bloor Research 2012 Confidential Bloor Research 2009 telling the Information Management story telling the right
  11. 11. Structured dataConfidential © © Bloor Research 2012 Confidential Bloor Research 2009 telling the Information Management story telling the right
  12. 12. Unstructured dataConfidential © © Bloor Research 2012 Confidential Bloor Research 2009 telling the Information Management story telling the right
  13. 13. Why - unstructuredConfidential © © Bloor Research 2012 Confidential Bloor Research 2009 telling the Information Management story telling the right
  14. 14. Why - structuredConfidential © © Bloor Research 2012 Confidential Bloor Research 2009 telling the Information Management story telling the right
  15. 15. How - Hadoop & NoSQL Key-value model: warehousing Column-family model: write once Document model: operationalConfidential © © Bloor Research 2012 Confidential Bloor Research 2009 telling the Information Management story telling the right
  16. 16. NoSQL vs Triple StoresConfidential © © Bloor Research 2012 Confidential Bloor Research 2009 telling the Information Management story telling the right
  17. 17. How – Hadoop basicsConfidential © © Bloor Research 2012 Confidential Bloor Research 2009 telling the Information Management story telling the right
  18. 18. MapReduceConfidential © © Bloor Research 2012 Confidential Bloor Research 2009 telling the Information Management story telling the right
  19. 19. But …Confidential © © Bloor Research 2012 Confidential Bloor Research 2009 telling the Information Management story telling the right
  20. 20. Alternatives/extensions to HadoopConfidential © © Bloor Research 2012 Confidential Bloor Research 2009 telling the Information Management story telling the right
  21. 21. Hadoop and the warehouse Technology Results and Applications 1 Join pre-computed DW and Hadoop Client side federation results DW For example, in telecoms you might use a data warehouse to provide aggregated customer loyalty data while Hadoop holds aggregated network utilisation data; and then a 3rd party product such as Quest Toad for Cloud can bring the data together from both sources, linking customer loyalty to network utilisation or network faults such as dropped calls.Confidential © © Bloor Research 2012 Confidential Bloor Research 2009 telling the Information Management story telling the right
  22. 22. Hadoop and Sybase IQ For example, in telecoms you might use a data warehouse to provide aggregated customer loyalty data while Hadoop holds aggregated network utilisation data; and then a 3rd party product such as Quest Toad for Cloud can bring the data together from both sources, linking customer loyalty to network utilisation or network faults such as dropped calls.Confidential © © Bloor Research 2012 Confidential Bloor Research 2009 telling the Information Management story telling the right
  23. 23. Hadoop and the warehouse Results and Applications Technology 2 Fetch HDFS subsets into DW warehouse ETL Hadoop data into DW Process entirely in warehouse An example use case is in eCommerce, where clickstream data from weblogs are stored in HDFS, and outputs of MapReduce jobs on that data (to study browsing behaviour) are loaded into the data warehouse via an ETL process. The transactional sales data in the warehouse is joined with the clickstream data to understand and predict customer browsing and buying behaviour.Confidential © © Bloor Research 2012 Confidential Bloor Research 2009 telling the Information Management story telling the right
  24. 24. Hadoop and Sybase IQ An example use case is in eCommerce, where clickstream data from weblogs are stored in HDFS, and outputs of MapReduce jobs on that data (to study browsing behaviour) are loaded into the data warehouse via an ETL process. The transactional sales data in the warehouse is joined with the clickstream data to understand and predict customer browsing and buying behaviour.Confidential © © Bloor Research 2012 Confidential Bloor Research 2009 telling the Information Management story telling the right
  25. 25. Hadoop and the warehouse Technology Results and Applications 3 C++/JAVA Fetch HDFS subsets by UDF materializing inside DW as in-memory Federate Hadoop data DW virtual tables into DW Files Triggered on the fly as part of query An example here would be in retail, where point of sale (POS) detailed data is stored in Hadoop. The warehouse fetches the POS data at fixed intervals from HDFS for specific hot selling SKUs, combined with inventory data from the warehouse to predict and prevent inventory “stockouts”.Confidential © © Bloor Research 2012 Confidential Bloor Research 2009 telling the Information Management story telling the right
  26. 26. Hadoop and Sybase IQ POS Data Inventory Data HDFS UDF Bridge Sybase IQ An example here would be in retail, where point of sale (POS) detailed data is stored in Hadoop. The warehouse fetches the POS data at fixed intervals from HDFS for specific hot selling SKUs, combined with inventory data from the warehouse to predict and prevent inventory “stockouts”.Confidential © © Bloor Research 2012 Confidential Bloor Research 2009 telling the Information Management story telling the right
  27. 27. Hadoop and the warehouse Technology Results and Applications C++/JAVA UDF 4 Trigger MapReduce job in both Federate Hadoop jobs DW and Hadoop and join the DW into DW results on the fly You might use this approach in the utility sector, with smart meter and smart grid data combined for load monitoring and demand forecasting. Smart grid transmission quality data (multi-attribute time series data) stored in HDFS can be computed via Hadoop MapReduce jobs triggered from the data warehouse and combined with smart meter data stored in the warehouse, to analyse demand and workload.Confidential © © Bloor Research 2012 Confidential Bloor Research 2009 telling the Information Management story telling the right
  28. 28. Hadoop and Sybase IQ Smart Grid Smart Meter Transmission Data Consumption Data HDFS UDF Bridge Sybase IQ You might use this approach in the utility sector, with smart meter and smart grid data combined for load monitoring and demand forecasting. Smart grid transmission quality data (multi-attribute time series data) stored in HDFS can be computed via Hadoop MapReduce jobs triggered from the data warehouse and combined with smart meter data stored in the warehouse, to analyse demand and workload.Confidential © © Bloor Research 2012 Confidential Bloor Research 2009 telling the Information Management story telling the right
  29. 29. Hadoop and Sybase IQ NoSQL scale out is easier and cheaper than scale up Warehouse is faster, more robust and is not limited to batch processing SQL programmers are less expensive and easier to get Time stamps may be an issueConfidential © © Bloor Research 2012 Confidential Bloor Research 2009 telling the Information Management story telling the right
  30. 30. Conclusion We can now analyse data in ways that were not previously possible that can provide new insights into customer (and other) behaviour We can do this cost-effectively But there are limitations that will typically require a combined warehouse/Hadoop environment There are various ways in which this can be achieved.Confidential © © Bloor Research 2012 Confidential Bloor Research 2009 telling the Information Management story telling the right
  31. 31. THANK YOU FOR MORE INFORMATION WWW.SYBASE.COM/SYBASEIQBIGDATA 31 – Company Confidential – March 27, 2012

×