Nationwide Insurance use case of big data using BigInsights

1. © 2015 IBM Corporation
From Concept to Production:
Nationwide Insurance IBM BigInsights
Journey with Telematics # 2404
Krish Rajaram & Rajesh Nandagiri – 10/26/2015

2. Big Data and
Analytics Helps
Nationwide
Customers Become
Better Drivers

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. Introduction
About Nationwide
SmartRide Program
SmartRide Data

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. 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. 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. 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. Architecture
Logical Data Flow
IBM® BigInsights™ Configuration
Decision Catalog
Job Orchestration

10. Logical Data Flow
9

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. Decision Catalog
11

13. Job Orchestration
12

14. Data Processing
Design Considerations
Phases of Data Movement
Batch Performance Metrics

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. 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. 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. 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. 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. 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. Data Access
SmartRide Web Page
Application Layer
Column Family and Row Key Design
Performance Metrics

22. SmartRide Web Page
21

23. SmartRide Web Page – Daily
22

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. 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. 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. HBase Data Distribution – Using Hannibal
26
SmartRide Data Distribution

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. 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. © 2015 IBM Corporation
Thank You

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. 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. 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. • 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