Think Big, A Teradata Company
Rick Farnell, Co-Founder & SVP International
2
Open Data Platform Support
© 2015 Think Big, a Teradata Company
3
What we learned over 5 years at Think Big
Teamwork is Critical Skills Matter Celebrate Success
© 2015 Think Big, a Teradata Company
4
Our hunch was right…this market will be BIG
Source: www.Indeed.com April 8, 2015
© 2015 Think Big, a Teradata Company
http://wikibon.org/wiki/v/Big_Data_Vendor_Revenue_and_Market_Forecast_2013-2017
5
• 100% Big Data Focus
• Founded in 2010 with over 100 engagements across 70 clients
• Unlock client value of big data with data science and data engineering services
• Proven vendor-neutral open source integration expertise
• Agile team-based development methodology
• Think Big Academy for skills and organizational development
• Global delivery model on-site services with near shore and off shore support
Who is Think Big?
© 2015 Think Big, a Teradata Company
6
Think Big Services Engagement Model
STRATEGY IMPLEMENTATION SOLUTION SUPPORT
Think Big offers end-to-end Big Data strategy, implementation and support services focused on
helping customers quickly achieve ROI on their Big Data investments
Enterprise Data Lake Software Frameworks
Big Data
Roadmap
Establish
Data Lake
Analytic
Solutions
Managed
Services
Data Lake
Optimization
© 2015 Think Big, a Teradata Company
7
5 Years of Big Data Services with Industry Leaders
Think Big Clients
eCommerce
2 of Global Top 5
Retail
2 of Global Top 5
Social Networking
Global #1
Banking
4 of Global Top 10
Credit Issuer
2 of Global Top 5
Financial Data Services
2 of Global Top 5
Financial Exchanges
Global #2
Brokerage & Mutual Funds
2 of Global Top 5
Asset Management
Global #1
Semiconductor
2 of Global Top 5
Data Storage Devices
3 of Global Top 5
Disk Drive Manufacturing
Global #1
Telecommunications
2 of Global Top 5
Media & Advertising
2 of Global Top 4
Internet Transaction Security
Global #1
© 2015 Think Big, a Teradata Company
8
Top 5 Learnings over 5 Years in Big Data
1. Big Data is a journey not an event.
2. Open source is great for innovation but requires engineering
sophistication in design patterns & best practices to continually scale
analytics in production.
3. Importance of metadata, lineage, data management, data
governance, data quality is critical to successful enterprise Big Data
deployments.
4. A new data platform will not resolve organizational problems.
5. The Hadoop & opensource ecosystem is moving at the speed of light,
you must be agile. Test, build, learn, repeat.
© 2015 Think Big, a Teradata Company
Think Big Enterprise Data Lake Case Study:
Global Manufacturing Client
10
Manufacturing Client Overview
Metrics:
• Collecting >2M manufacturing/testing binary files daily across Americas and APAC Facilities
• Collecting from ~500 tables across 6 databases  tens of millions of records daily
• Over 140 analytics users to date
• Over 150 attendees participated in Big Data Platform training
Business Goals for Investment in Enterprise Data Lake:
• Provide Enterprise-wide data access for timely analytics to product quality engineers
• Establish foundation for large scale proactive manufacturing and quality analytics
• Reduce $Millions in costs of scrap waste
• Reduce $Millions in costs of containment processes
• Reduce $Millions in costs of data search parties
• Increase revenue and market share by accelerating time to market of products
© 2015 Think Big, a Teradata Company
11
Engineering and Ingestion Design are Critical Day 1
1. Ingestion1. Utilize a robust Ingestion
Design with a Buffer Server
Zone
2. Packaging
2. Robust packaging many
small flies into larger files
with metadata.
3. Parse on Demand
3. Parse on demand.
Increase velocity with right
level of parsing, parse and
refine in more detail as
needed.
4. Establish Zones for Control
4. Establish Zones for
comprehensive pipeline
control, buffer, landing,
ingest, core and publish.
© 2015 Think Big, a Teradata Company
12
Think Globally
EDW
EDW
5. Replication on Ingestion
5. Replicate at ingestion
with collection of key
metadata information.
6. Data Treatment Facility
6. Utilize HDFS as giant file
system, all data lands but
some will continue in pipe
to other MPP/EDW systems
etc.
7. WAN Accelerators
7. WAN accelerates the
speed transfer of ASIA
to/from North America
8. Published Data
8. Processing and format is
customized for the
audience and
consumption pattern.
© 2015 Think Big, a Teradata Company
13
Governance for Production Workloads
9. Design Zones for Enterprise Governance
9. Zones will have different
retention periods and
different access patterns
and in order to have
pipeline control within your
governance strategy you
must design for this upfront.
© 2015 Think Big, a Teradata Company
14
Optimize Access Patterns of your Data in Hadoop
10. Relational access pattern
10. Optimized for known
queries. Equivalent of a
Ferrari moving a bag of
groceries.
11. Hadoop access pattern
11. Recast your problem.
Hadoop optimal for large
batches of work. Optimize
hierarchical queries with
transitive closure.
Equivalent of a freight train
moving multiple
warehouses of groceries.
© 2015 Think Big, a Teradata Company
15
Pipeline Management, Monitoring & Control
HDFS
12. Data Pipeline
12. End to End Pipeline
visibility is extremely
important in an Enterprise
Data Lake. Metadata
Design and use of best
practices are key in the use
of this pattern.
© 2015 Think Big, a Teradata Company
16
Detailed Ingestion Design Patterns
13. Source Systems
13. Hundreds of servers &
hundreds of data streams
require expert enterprise
engineering.
14. Utilities
14. Layer in logging,
monitoring and scheduling
for complete control of
your Enterprise Data Lake.
© 2015 Think Big, a Teradata Company
17
ROI for Enterprise Data Lake Manufacturing Client
Est. at $11M year 1, $30M cumulative year 2
© 2015 Think Big, a Teradata Company
Operations Engineer: a recent production issue required detailed historical testing data. Our current
systems did not have the required retention for this request. The Big Data team was able to pull and
analyze all the required data from the Big Data Platform in minutes, as opposed to 3+ weeks that we
used to take to pull the data from multiple systems and off tape archive.
Legacy Systems Enterprise Data Lake
Retention 3-6 months scattered
DBs & tape archive.
100% data online in
enterprise data lake for 3+
years.
Coverage Summaries, samples
for several data sets.
100% parametric data
captured in raw form.
Analysis Reactive, missed
quality improvement
opportunities.
Daily dashboard with larger
data sets to support
proactive improvements.
18 © 2015 Think Big, a Teradata Company
Enterprise Data Lake
Information Sources
Evaluate
Source Data
Ingest
Collect & Manage
Metadata
Apply
Structure
Sequence
Compress
Automate
Protect
Prepare Data
for Ingest
Prepare Source
Metadata
Perimeter-Authentication-Authorization
InfoSec
Downstream
Applications
Dashboard
Engine
Think Big Enterprise Data Lake
Industry Leading Assets, Services and Methodology
19
20 © 2015 Think Big, a Teradata Company
• Think Big is expanding, bringing its focus
on open source consulting to the
international region
• An office in the UK’s London Bridge
Business district will serve as its
international hub
• Think Big is aggressively hiring a team of
data engineers, data scientists,
technology project managers and sales
leaders
• Rick Farnell, Think Big co-founder and
SVP, International will lead the
international practice
London Bridge Business District
Photo credit: Duncan Harris. Courtesy of Flickr. Creative Commons
Think Big International Expansion
21 © 2015 Think Big, a Teradata Company
Dublin, Ireland Munich, Germany Mumbai, India
Think Big International Expansion: Phase 1
Think Big International Hub
London, England
Photo credits: London (Duncan Harris). Dublin (Guiseppe Milo), Munich (John Morgan), Mumbai (McKay Savage).
Courtesy of Flickr. Creative Commons
22
23
Dashboard Engine for Hadoop
Fast Access for Comprehensive Historical Data Stored in Hadoop
“Right” time data
Latencies under a second
Scales easily for thousands of simultaneous users
Reporting, Visualization and
Analytics
© 2015 Think Big, a Teradata Company
Visit our Think Big Booth
to see our Dashboard Engine Demo.
Enterprise Data Lake
Information Sources
Downstream
Applications
24 © 2015 Think Big, a Teradata Company
Thank you
Rick Farnell
Co-Founder and SVP International, Think Big
Rick.Farnell@thinkbiganalytics.com

Hadoop 2015: what we larned -Think Big, A Teradata Company

  • 1.
    Think Big, ATeradata Company Rick Farnell, Co-Founder & SVP International
  • 2.
    2 Open Data PlatformSupport © 2015 Think Big, a Teradata Company
  • 3.
    3 What we learnedover 5 years at Think Big Teamwork is Critical Skills Matter Celebrate Success © 2015 Think Big, a Teradata Company
  • 4.
    4 Our hunch wasright…this market will be BIG Source: www.Indeed.com April 8, 2015 © 2015 Think Big, a Teradata Company http://wikibon.org/wiki/v/Big_Data_Vendor_Revenue_and_Market_Forecast_2013-2017
  • 5.
    5 • 100% BigData Focus • Founded in 2010 with over 100 engagements across 70 clients • Unlock client value of big data with data science and data engineering services • Proven vendor-neutral open source integration expertise • Agile team-based development methodology • Think Big Academy for skills and organizational development • Global delivery model on-site services with near shore and off shore support Who is Think Big? © 2015 Think Big, a Teradata Company
  • 6.
    6 Think Big ServicesEngagement Model STRATEGY IMPLEMENTATION SOLUTION SUPPORT Think Big offers end-to-end Big Data strategy, implementation and support services focused on helping customers quickly achieve ROI on their Big Data investments Enterprise Data Lake Software Frameworks Big Data Roadmap Establish Data Lake Analytic Solutions Managed Services Data Lake Optimization © 2015 Think Big, a Teradata Company
  • 7.
    7 5 Years ofBig Data Services with Industry Leaders Think Big Clients eCommerce 2 of Global Top 5 Retail 2 of Global Top 5 Social Networking Global #1 Banking 4 of Global Top 10 Credit Issuer 2 of Global Top 5 Financial Data Services 2 of Global Top 5 Financial Exchanges Global #2 Brokerage & Mutual Funds 2 of Global Top 5 Asset Management Global #1 Semiconductor 2 of Global Top 5 Data Storage Devices 3 of Global Top 5 Disk Drive Manufacturing Global #1 Telecommunications 2 of Global Top 5 Media & Advertising 2 of Global Top 4 Internet Transaction Security Global #1 © 2015 Think Big, a Teradata Company
  • 8.
    8 Top 5 Learningsover 5 Years in Big Data 1. Big Data is a journey not an event. 2. Open source is great for innovation but requires engineering sophistication in design patterns & best practices to continually scale analytics in production. 3. Importance of metadata, lineage, data management, data governance, data quality is critical to successful enterprise Big Data deployments. 4. A new data platform will not resolve organizational problems. 5. The Hadoop & opensource ecosystem is moving at the speed of light, you must be agile. Test, build, learn, repeat. © 2015 Think Big, a Teradata Company
  • 9.
    Think Big EnterpriseData Lake Case Study: Global Manufacturing Client
  • 10.
    10 Manufacturing Client Overview Metrics: •Collecting >2M manufacturing/testing binary files daily across Americas and APAC Facilities • Collecting from ~500 tables across 6 databases  tens of millions of records daily • Over 140 analytics users to date • Over 150 attendees participated in Big Data Platform training Business Goals for Investment in Enterprise Data Lake: • Provide Enterprise-wide data access for timely analytics to product quality engineers • Establish foundation for large scale proactive manufacturing and quality analytics • Reduce $Millions in costs of scrap waste • Reduce $Millions in costs of containment processes • Reduce $Millions in costs of data search parties • Increase revenue and market share by accelerating time to market of products © 2015 Think Big, a Teradata Company
  • 11.
    11 Engineering and IngestionDesign are Critical Day 1 1. Ingestion1. Utilize a robust Ingestion Design with a Buffer Server Zone 2. Packaging 2. Robust packaging many small flies into larger files with metadata. 3. Parse on Demand 3. Parse on demand. Increase velocity with right level of parsing, parse and refine in more detail as needed. 4. Establish Zones for Control 4. Establish Zones for comprehensive pipeline control, buffer, landing, ingest, core and publish. © 2015 Think Big, a Teradata Company
  • 12.
    12 Think Globally EDW EDW 5. Replicationon Ingestion 5. Replicate at ingestion with collection of key metadata information. 6. Data Treatment Facility 6. Utilize HDFS as giant file system, all data lands but some will continue in pipe to other MPP/EDW systems etc. 7. WAN Accelerators 7. WAN accelerates the speed transfer of ASIA to/from North America 8. Published Data 8. Processing and format is customized for the audience and consumption pattern. © 2015 Think Big, a Teradata Company
  • 13.
    13 Governance for ProductionWorkloads 9. Design Zones for Enterprise Governance 9. Zones will have different retention periods and different access patterns and in order to have pipeline control within your governance strategy you must design for this upfront. © 2015 Think Big, a Teradata Company
  • 14.
    14 Optimize Access Patternsof your Data in Hadoop 10. Relational access pattern 10. Optimized for known queries. Equivalent of a Ferrari moving a bag of groceries. 11. Hadoop access pattern 11. Recast your problem. Hadoop optimal for large batches of work. Optimize hierarchical queries with transitive closure. Equivalent of a freight train moving multiple warehouses of groceries. © 2015 Think Big, a Teradata Company
  • 15.
    15 Pipeline Management, Monitoring& Control HDFS 12. Data Pipeline 12. End to End Pipeline visibility is extremely important in an Enterprise Data Lake. Metadata Design and use of best practices are key in the use of this pattern. © 2015 Think Big, a Teradata Company
  • 16.
    16 Detailed Ingestion DesignPatterns 13. Source Systems 13. Hundreds of servers & hundreds of data streams require expert enterprise engineering. 14. Utilities 14. Layer in logging, monitoring and scheduling for complete control of your Enterprise Data Lake. © 2015 Think Big, a Teradata Company
  • 17.
    17 ROI for EnterpriseData Lake Manufacturing Client Est. at $11M year 1, $30M cumulative year 2 © 2015 Think Big, a Teradata Company Operations Engineer: a recent production issue required detailed historical testing data. Our current systems did not have the required retention for this request. The Big Data team was able to pull and analyze all the required data from the Big Data Platform in minutes, as opposed to 3+ weeks that we used to take to pull the data from multiple systems and off tape archive. Legacy Systems Enterprise Data Lake Retention 3-6 months scattered DBs & tape archive. 100% data online in enterprise data lake for 3+ years. Coverage Summaries, samples for several data sets. 100% parametric data captured in raw form. Analysis Reactive, missed quality improvement opportunities. Daily dashboard with larger data sets to support proactive improvements.
  • 18.
    18 © 2015Think Big, a Teradata Company Enterprise Data Lake Information Sources Evaluate Source Data Ingest Collect & Manage Metadata Apply Structure Sequence Compress Automate Protect Prepare Data for Ingest Prepare Source Metadata Perimeter-Authentication-Authorization InfoSec Downstream Applications Dashboard Engine Think Big Enterprise Data Lake Industry Leading Assets, Services and Methodology
  • 19.
  • 20.
    20 © 2015Think Big, a Teradata Company • Think Big is expanding, bringing its focus on open source consulting to the international region • An office in the UK’s London Bridge Business district will serve as its international hub • Think Big is aggressively hiring a team of data engineers, data scientists, technology project managers and sales leaders • Rick Farnell, Think Big co-founder and SVP, International will lead the international practice London Bridge Business District Photo credit: Duncan Harris. Courtesy of Flickr. Creative Commons Think Big International Expansion
  • 21.
    21 © 2015Think Big, a Teradata Company Dublin, Ireland Munich, Germany Mumbai, India Think Big International Expansion: Phase 1 Think Big International Hub London, England Photo credits: London (Duncan Harris). Dublin (Guiseppe Milo), Munich (John Morgan), Mumbai (McKay Savage). Courtesy of Flickr. Creative Commons
  • 22.
  • 23.
    23 Dashboard Engine forHadoop Fast Access for Comprehensive Historical Data Stored in Hadoop “Right” time data Latencies under a second Scales easily for thousands of simultaneous users Reporting, Visualization and Analytics © 2015 Think Big, a Teradata Company Visit our Think Big Booth to see our Dashboard Engine Demo. Enterprise Data Lake Information Sources Downstream Applications
  • 24.
    24 © 2015Think Big, a Teradata Company Thank you Rick Farnell Co-Founder and SVP International, Think Big Rick.Farnell@thinkbiganalytics.com

Editor's Notes

  • #22 Think Big Americas Office Locations Mountain View Chicago Salt Lake City Boston New York