Successfully reported this slideshow.
We use your LinkedIn profile and activity data to personalize ads and to show you more relevant ads. You can change your ad preferences anytime.

Concept to production Nationwide Insurance BigInsights Journey with Telematics

2,722 views

Published on

Nationwide Insurance use case of big data using BigInsights

Published in: Data & Analytics

Concept to production Nationwide Insurance BigInsights Journey with Telematics

  1. 1. © 2015 IBM Corporation From Concept to Production: Nationwide Insurance IBM BigInsights Journey with Telematics # 2404 Krish Rajaram & Rajesh Nandagiri – 10/26/2015
  2. 2. Big Data and Analytics Helps Nationwide Customers Become Better Drivers
  3. 3. Agenda Introduction Architecture Data Processing Data Access Business Benefits 2 0.67 8.30 8.30 0.00 1.00 2.00 3.00 4.00 5.00 6.00 7.00 8.00 9.00 0.00 100.00 200.00 300.00 400.00 500.00 600.00 700.00 1 Hr Batch 12 Hrs Loop 12 Hrs batch DataVolumeinGB ElapseTimeinMins Cycles AfterRedesign First Iteration
  4. 4. Introduction About Nationwide SmartRide Program SmartRide Data
  5. 5. About Nationwide 4 16+ MILLION POLICIES 25MILLION CONTRIBUTED TO NONPROFITS AND COMMUNITIES $ 1# INSURER OF FARMS AND RANCHES 7LARGEST HOMEOWNER AND AUTO INSURANCE PROVIDER IN THE U.S. th GALLUP GREAT PLACE TO WORK AWARD WINNER 3 YEARS RUNNING LARGEST PET INSURER IN THE U.S. 9th LARGEST COMMERCIAL INSURER $23.9 BILLION IN REVENUE FOR 2013 Nationwide has approximately 31,000 associates serving customers in nearly every state. 1# PROVIDER OF PUBLIC-SECTOR RETIREMENT PLANS FOUNDED IN 1926 BY MEMBERS OF THE OHIO FARM BUREAU 28th COMPUTERWORLD GREAT PLACE TO WORK IN IT
  6. 6. About SmartRide • SmartRide is Nationwide's version of Telematics, offered to customers to help them improve their driving behavior and save on insurance premiums. 5 • Customers install a small device into their vehicle for 6 months which measures…
  7. 7. SmartRide Data Characteristics  Multiple vendors  Files of different layouts arriving at different frequencies:  Hourly  Every 4 hrs  Four CSV files per vendor  ~ 30 GB to ~ 60 GB of data per day  Data challenges  Late arriving trips  Partial trips  Duplicate trips  Orphan trips 6
  8. 8. Trip Data Characteristics • Missing Timestamp & Speed Spike • Acceleration Lag 7 vin_nb trip_nb position_ts Speed engine_rpm abc 123 2015-07-21 12:31:36.0 54 1600 abc 123 2015-07-21 12:31:39.0 55 1800 abc 123 2015-07-21 12:31:42.0 57 1500 abc 123 2015-07-21 12:31:43.0 82 1600 abc 123 2015-07-21 12:31:44.0 58 1500 vin_nb trip_nb position_ts Speed engine_rpm abc 123 2015-06-30 21:25:05.0 0 700 abc 123 2015-06-30 21:25:06.0 0 700 abc 123 2015-06-30 21:25:07.0 0 1000 abc 123 2015-06-30 21:25:08.0 8 1800 abc 123 2015-06-30 21:25:09.0 15 2000
  9. 9. Architecture Logical Data Flow IBM® BigInsights™ Configuration Decision Catalog Job Orchestration
  10. 10. Logical Data Flow 9
  11. 11. IBM® BigInsights™ for Apache™ Hadoop Configuration • Version 2.1.2  6 Management Nodes and 16 Data Nodes  Each with 128 GB RAM and 18 TB of storage  Hadoop 2.2, BigSQL 1.0, Hive 0.12, Hbase 0.96 • Three environments  Dev, Test, and Production All same configuration • Limitations  No workload management  No environment for DR  Used Test Cluster for Hbase failover 10
  12. 12. Decision Catalog 11
  13. 13. Job Orchestration 12
  14. 14. Data Processing Design Considerations Phases of Data Movement Batch Performance Metrics
  15. 15. Design Considerations • One hour window for end to end processing  Handling data issues  Summarization  Multiple cycles per day • Predictable run time for backlog processing when jobs fail • Reloading incorrect batch • Restart failed batch 14 0.67 8.30 8.30 0.00 1.00 2.00 3.00 4.00 5.00 6.00 7.00 8.00 9.00 0.00 100.00 200.00 300.00 400.00 500.00 600.00 700.00 1 Hr Batch 12 Hrs Loop 12 Hrs batch ElapseTimeinMins Cycles TripScrub TripSecDetails TripSum State Median Hbase load AuditEn raw2can Input Size in GB
  16. 16. Acquire Phase 15  Raw trip files copied into HDFS using WebHDFS protocol  Folders created by vendor, file, load event ID, and batch #  Used Sqoop to transfer 4 TB of historical data from Data Warehouse  Hive external tables for each file  Partitioned by load event ID and batch #  Used both BigSQL 1.0 and HiveQL  Partitioned external tables helped in o Processing backlog data o Reprocessing incorrect batches
  17. 17. Standardize Phase 16  Select data from external tables based on load event ID  Each load event ID can include one or more batches  More than one load event ID can be processed in one cycle  Data moved to next stage only from work tables  Helped in performance  Dynamic partitions helped in loading multiple batches  Partitions get overwritten if already exists  Helped in reprocessing incorrect batch Work tables contain data for CURRENT processing cycle Canonical tables partitioned by source and batch # Load using dynamic partitioning
  18. 18. Data Scrubbing & Event Calculation 17 Trip summary Trip point  Map side join  Single read multi write Orphan trips Trip points (Work table)  Java M/R program for o Scrubbing o Events calculation  Night time driving  Hard brake  Fast acceleration  Miles driven Events at seconds level (Work table)  Very good performance gain  Using Java for complex scrubbing rules  Single read multiple writes  Only required data points processed  No data persisted to corpse tables
  19. 19. Summarization Phase 18 Events at seconds level (Work table)  Gather all trips related to devices from current trip and aggregate at various levels  Union ALL  UDF to store data points for trip graph  Replace new summary info into final table SRE summary (Work table) SRE summary partitioned by source SRE summary in HBase  Parallelized the Union All operation  Partitioning by Source enabled both Vendor data to be processed at same time if overlap happens  PUT from Hive to Hbase, WAL disabled  Shorten column names  Changed to epoch time  Prefix salting key  Generate rowkey  Column family mapping
  20. 20. Batch Performance Metrics 19 1 Hr Batch SLA 0 0.5 1 1.5 2 2.5 0 5 10 15 20 25 30 35 40 DataSizeinGB RuntimeinMins Cycle Schedule Time Avg Run Times for Hourly Cycles 0 2 4 6 8 10 12 0 10 20 30 40 50 60 0000 0400 0800 1200 1600 2000 DataSizeinGB RuntimeinMins Cycle Schedule Time Avg Run Times for 4 hr Cycles Trip Second Details Standarize SRE Trip Summary Hive SRE Trip Summary Hbase Audits Acquire Size in GB
  21. 21. Data Access SmartRide Web Page Application Layer Column Family and Row Key Design Performance Metrics
  22. 22. SmartRide Web Page 21
  23. 23. SmartRide Web Page – Daily 22
  24. 24. Application Layer 23 Data Access Layer HBASE API Restful Service Single Page Web App BigSQL & Hive Aggregates Daily HDFS HBase HRegion Server HRegion HLog Memstore HFile ODS – DB2 ODBC
  25. 25. Column Family & RowKey Design 24 RowKey – Pfx_pgmId_pdflg_ timestamp Column Family – Summary Data Column Family – Trip-point Data 12_8798782_Tp_201 5080912000000 SM:miles,1500001245,’15’, SM:hb,1500001245,’2’, SM:fa,1500001245,’5’, SM:nt,1500001245,’Y’ TP:Trip,1500001245,’{JSON BLOB}’ Sorted Lexicographically • Column family (CF) helps in grouping the related columns depending on access pattern. • Co-locating the keys related to one customer in one region to access data using filter from one region server.
  26. 26. Performance Metrics Scenarios – 1x, 2x, 3x concurrent users, Zookeeper node going down, Datanode unavailable Tools used – Initial test using custom program, LoadRunner for final test, SiteScope for monitoring resource consumption 25 SLA for aggregates – 5 sec # of concurrent users - 1200
  27. 27. HBase Data Distribution – Using Hannibal 26 SmartRide Data Distribution
  28. 28. Business Benefits • Deeper Engagement with Members  Over 2 million website page views since the July launch. To put in perspective, our vendor-hosted website would receive 100,000 views in a 12 month period.  Over 60K users have accessed the new site and 90% of those are new users. • Increase in bind ratios across all channels • Improvement in loss ratios • Enterprise first "big data" implementation at Nationwide 27
  29. 29. Future scope – Personal and Commercial Fleet 28 Insights Give Nationwide Competitive Advantage Weather Data GPS Data Hourly Trip Data from Device Claims Data Other Public Records
  30. 30. © 2015 IBM Corporation Thank You
  31. 31. We Value Your Feedback! Don’t forget to submit your Insight session and speaker feedback! Your feedback is very important to us – we use it to continually improve the conference. Access your surveys at insight2015survey.com to quickly submit your surveys from your smartphone, laptop, or conference kiosk. 30
  32. 32. 31 Notices and Disclaimers Copyright © 2015 by International Business Machines Corporation (IBM). No part of this document may be reproduced or transmitted in any form without written permission from IBM. U.S. Government Users Restricted Rights - Use, duplication or disclosure restricted by GSA ADP Schedule Contract with IBM. Information in these presentations (including information relating to products that have not yet been announced by IBM) has been reviewed for accuracy as of the date of initial publication and could include unintentional technical or typographical errors. IBM shall have no responsibility to update this information. THIS DOCUMENT IS DISTRIBUTED "AS IS" WITHOUT ANY WARRANTY, EITHER EXPRESS OR IMPLIED. IN NO EVENT SHALL IBM BE LIABLE FOR ANY DAMAGE ARISING FROM THE USE OF THIS INFORMATION, INCLUDING BUT NOT LIMITED TO, LOSS OF DATA, BUSINESS INTERRUPTION, LOSS OF PROFIT OR LOSS OF OPPORTUNITY. IBM products and services are warranted according to the terms and conditions of the agreements under which they are provided. Any statements regarding IBM's future direction, intent or product plans are subject to change or withdrawal without notice. Performance data contained herein was generally obtained in a controlled, isolated environments. Customer examples are presented as illustrations of how those customers have used IBM products and the results they may have achieved. Actual performance, cost, savings or other results in other operating environments may vary. References in this document to IBM products, programs, or services does not imply that IBM intends to make such products, programs or services available in all countries in which IBM operates or does business. Workshops, sessions and associated materials may have been prepared by independent session speakers, and do not necessarily reflect the views of IBM. All materials and discussions are provided for informational purposes only, and are neither intended to, nor shall constitute legal or other guidance or advice to any individual participant or their specific situation. It is the customer’s responsibility to insure its own compliance with legal requirements and to obtain advice of competent legal counsel as to the identification and interpretation of any relevant laws and regulatory requirements that may affect the customer’s business and any actions the customer may need to take to comply with such laws. IBM does not provide legal advice or represent or warrant that its services or products will ensure that the customer is in compliance with any law.
  33. 33. 32 Notices and Disclaimers (con’t) Information concerning non-IBM products was obtained from the suppliers of those products, their published announcements or other publicly available sources. IBM has not tested those products in connection with this publication and cannot confirm the accuracy of performance, compatibility or any other claims related to non-IBM products. Questions on the capabilities of non-IBM products should be addressed to the suppliers of those products. IBM does not warrant the quality of any third-party products, or the ability of any such third-party products to interoperate with IBM’s products. IBM EXPRESSLY DISCLAIMS ALL WARRANTIES, EXPRESSED OR IMPLIED, INCLUDING BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE. The provision of the information contained herein is not intended to, and does not, grant any right or license under any IBM patents, copyrights, trademarks or other intellectual property right. • IBM, the IBM logo, ibm.com, Aspera®, Bluemix, Blueworks Live, CICS, Clearcase, Cognos®, DOORS®, Emptoris®, Enterprise Document Management System™, FASP®, FileNet®, Global Business Services ®, Global Technology Services ®, IBM ExperienceOne™, IBM SmartCloud®, IBM Social Business®, Information on Demand, ILOG, Maximo®, MQIntegrator®, MQSeries®, Netcool®, OMEGAMON, OpenPower, PureAnalytics™, PureApplication®, pureCluster™, PureCoverage®, PureData®, PureExperience®, PureFlex®, pureQuery®, pureScale®, PureSystems®, QRadar®, Rational®, Rhapsody®, Smarter Commerce®, SoDA, SPSS, Sterling Commerce®, StoredIQ, Tealeaf®, Tivoli®, Trusteer®, Unica®, urban{code}®, Watson, WebSphere®, Worklight®, X-Force® and System z® Z/OS, are trademarks of International Business Machines Corporation, registered in many jurisdictions worldwide. Other product and service names might be trademarks of IBM or other companies. A current list of IBM trademarks is available on the Web at "Copyright and trademark information" at: www.ibm.com/legal/copytrade.shtml.
  34. 34. • IBM’s statements regarding its plans, directions, and intent are subject to change or withdrawal without notice at IBM’s sole discretion. • Information regarding potential future products is intended to outline our general product direction and it should not be relied on in making a purchasing decision. • The information mentioned regarding potential future products is not a commitment, promise, or legal obligation to deliver any material, code or functionality. Information about potential future products may not be incorporated into any contract. • The development, release, and timing of any future features or functionality described for our products remains at our sole discretion. Performance is based on measurements and projections using standard IBM benchmarks in a controlled environment. The actual throughput or performance that any user will experience will vary depending upon many factors, including considerations such as the amount of multiprogramming in the user’s job stream, the I/O configuration, the storage configuration, and the workload processed. Therefore, no assurance can be given that an individual user will achieve results similar to those stated here. Please Note: 2

×