going from big data to big answers - files.meetup.comfiles.meetup.com/10136492/intel big data use...
TRANSCRIPT
Intel Information Technology
Going From Big Data to Big AnswersApril 17, 2014
Ajay Chandramouly
@ajayc47
Intel Information Technology
Agenda
• Impact and Value of Big Data
• Intel IT Use Cases – Bringing Value of Big Data to Intel
• Call to Action – Bringing Value of Big Data to Your Business
2
Intel Information Technology
Big
Data
TEXT
Intel Information Technology
Big Data AnalyticsValue
= the “Asset” = the “Action”
Intel Information Technology
The Four Pillars Of Big Data
5
Volume
Massive scale and growth of unstructured data
• 80%~90% of total data
• Growing 10x~50x faster than structured (relational) data
• 10x~100x of traditional data warehousing
Velocity
Real-time rather than batch-style analysis
• Data streamed in, tortured, and discarded
• Making impact on the spot rather than after-the-fact
Variety
Heterogeneity and variable nature of Big Data
• Many different forms (text, document, image, video, ...)
• No schema or weak schema
• Inconsistent syntax and semantics
Variability
Predictive analytics for future trends and patterns
• Deep, complex analysis (machine learning, statistic modeling, graph algorithms, …), versus
• Traditional business intelligence (querying, reporting, …)
Big Data augments traditional Business Intelligence
Intel Information Technology Intel Information Technology
BIG DATAMACHINE
GENERATED
HUMANGENERATED
BUSINESSGENERATED
Edge
Scale Up
Distributed
REQUIRES DIFFERENT APPROACHES
Scale Out
NETWO
RK
STORAGECOMPUTE
Intel® Optimized Big Data
In Memory
XDW
MPP
One Size Doesn’t Fit All
LOB
IOT
Intel Information Technology Intel Information Technology
The Fusion of Edge Devices and Big Data Analytics
Planogram monitoring(Real-time stock level)
Interactive display(Behavioral marketing)
Real-time transaction
Store heat map (hot merchandise, browsing history, conversion rate)
RFID
RFID
Dynamic pricingReal-time, personalized AdAuto promotion/couponSocial network connections
Surveillance camera (store statistics)
Intel Information Technology 8
Going from Data to Actionable Insight
Intel Information Technology Intel Information Technology
Intel IT – What We’re Doing in Big Data
9
Intel Information Technology Intel Information Technology
6,500 IT Employees59 IT sites globally
150,000 Connected Systems40,000 Handheld Devices
100,000 Intel Employees164 Intel Sites across 63 Countries
68 Data Centers25% reduction with virtualization
Inspire employees
IT is business
Changing traditional thinking
Service reliability
Intel Confidential
10
Intel Information Technology Intel Information Technology 11
Our IT Environment - Continued
Intel Information Technology Intel Information Technology
IT Leadership
12
Transform
Contribute Value
Deliver Services
“License to Decide”
Strategic Relationship
“Right to Influence”
Collaborative Relationship
“Reason to Exist”
Transactional Relationship
Advanced Analytics Enables IT to Transform the Business
Intel Information Technology
Intel IT Vision for Big Data Analytics
13
Priority
We run big data analytics programs in each of our key lines of businesses. Also, all our key strategic initiatives have a big data component.
Strategy
Implement an internal, cost-effective big data platform
and in- parallel build the necessary skill set within
the organization,
Approach
Gradually build business value through advanced
analytics of big data.
Business Value
The value of our big data efforts was about USD
$100M in 2013. We expect that figure to grow in 2014.
IT formed an enterprise Big Data Analytics organization which solves High Value problems
Intel Information Technology Intel Information Technology
Big Data Path to Competitive Advantage
14
SalesWeb usage data
for
Marketing/Camp
aign predictions
(What)
New Biz ITIT Incident
Predictability
Context Aware
Analytics for
LBS
SecurityNetwork
Intrusion
Prediction and
Prevention
Big Data Use Cases
Tailor-made and Unique Big Data environment based on Intel needs
2011 - 12
•Defined strategy and implementation plan
•Hadoop Path-finding
•Deployed chosen MPP platform
•Acquired big data skills
•Deployed 3 big data projects (3 done)
•Completed big data distribution evaluation
•Landed internal Hadoop cluster in Prod
• Implemented Internal Hadoop Production cluster
2013
• Implement Internal Hadoop Pre-Production cluster
•Deliver a solid platform for the first set of use cases.
•Deploy internal 5+6=10 projects on top of the BI big data platform
•Deploy the qualified Big Data business use cases
•Deliver business value with this platform through the use of it.
•Expand Big Data Platforms to support use case demand.
• Setup BDP as a service with integration of IT processes.
•Prescriptive guidance for development and architecture.
•Standardize processes & tools
2014 • Expand IBD platform for the next set of
use cases. Deliver business value through the use of it.
Deploy internal 5+6=100 projects on top of the BI big data platform
• Evolve the IBD platform towards the next generation Hadoop ecosystem
Adopt IDH3 with Hadoop 2.0/YARN
Hbase for storage intensive use cases
Explore SQL on Hadoop use cases
Expand Big Data Platforms to support Enterprise BI use cases.
•Continuous improvement and expansion of platform, capabilities, guidance, process and tools.
MfgAsses feasibility
of Hadoop for
MIDAS as lower
cost solution
HR - POCTalent
Intelligence
Intel Information Technology Intel Information Technology
Intel IT Multi Data Warehouse Strategy
15
Big Data is a Part of a Comprehensive BI Strategy
Intel Information Technology
Intel IT’s Big Data Platform
MPP Platform
3rd-party solution
100x faster than traditional systems
Intel® Xeon® processor E7 family blades scale easily
Intel Data Platform
Based on Apache Hadoop
Optimized for Intel® Xeon processors,
SSD and 10GbE
HBase NoSql DB
Spark (In-Memory Analytics)
MPP – Massively Parallel Processing
Predictive Analytics Engine In house development
Enables real time, on-going Predictive service
Intel® Xeon® processor E7 family
Intel Data Platform
Intel Information Technology
Hadoop Use CasesContextual Recommendation Engine: Provides
recommendation engine and analytic capabilities to
acquisition.
Value:
•Provides new, intelligent capabilities and map
management technologies which can be offered as paid
services
Incident Predictability: Reduces incidents, impact on users and IT
Value:
•Reduces number of new incidents and employee
support costs while boosting productivity
Web Data Mining & Customer Insight: Provides customer and network usage analytics for
Intel.com and customer advertising
Value:
•Provides means to predict and adjust product position
or pricing based on response to marketing campaigns
Intel Information Technology Intel Information Technology
Contextual Recommendation Engine
Telmap Server
Offer Manager
Request for Navigation
Request for Available offers
Online –Model Inquiry
Offline -Model
Creation
Context Aware Recommendation System
Request for Offers Ranking based on User Preferences
Set of Offers for Display
Personal, context aware location based recommendations system to enrich user experience
Embed into Telmap navigation application
Provide commercial recommendations to users based on uncovered preferences
Recommendations are time, location and context aware based
Big Data Solution, Generic recommendation engine
5/5/2014
Intel Information Technology Intel Information Technology
Mobile Application Screen Shot
19
Intel Information Technology Intel Information Technology
Events Incidents
Symptom
table
More than 80% of event data is grouped
using only 10 event IDs .
Selection of critical fields by event
ID to facilitate Problem Managers
look into data
This is a way to look
into events we didn’t
have before… [Intel IT Problem
Manager]
Symptoms are based on group
of events.
- Volume- Frequency/day- Machine Importance- Application
Importance- Symptom effect- N. Incidents
Machine crying for
help (VOM)
Humans crying for
help (VOC)vs
Important events
bubble up with
higher Impact
Linear regression, Time Series Analysis and Predict
future Impact
Symptom Impact will
allow PMs to focus on
critical events requiring
help.
Better understanding of potential future problems since
Events data ALWAYS arrive first than Incident data.
Linear regression to predict future behavior
Event Data Incident Data
Using VOM (Voice of Machines) THEN will
help to PREDICT now VOC (Voice of
Customers)
- Tokenization- Stop-words- Steaming- Synonyms
Incident Predictability Big Picture
Intel Information Technology Intel Information Technology
Customer Insight: Web AnalyticsPurpose of PoC• Prove landing and ingesting of Web Data in Hadoop.
• Determining & landing of processed data in right container
• Get learnings for Path-to-production in terms of architecture options and nuances associated with Hadoop.
Business Value• Customer Insight owns delivery of the “Profile 330” and Analytics to support SMG• ROI: $20M Smart Analytics; $10M Demand Generation
5/5/2014
Intel Information Technology Intel Information Technology
Call to Action
22
Intel Information Technology
23
People & Skills
CxO Program Manager Project Manager Solutions Architect Data Architect Data Engineer SE/Developer DBA
Intel Information Technology
Intel’s Datacenter Portfolio
24
• Intel® Atom® & Xeon® storage solutions•SSD’s with Cache Acceleration•Luster•Next Gen NVM
•Open Network Platforms•Wind River OS•Quick Assist Technology•Silicon Photonics•Ethernet Switch & NIC Si
• Intel® Xeon® Product Family E3-E5-E7• Intel® Atom™• Intel® Xeon Phi™• Integrated graphics•TXT•AESNI
Storage Network Compute Optimized, Scalable,
Efficient Architecture
Platform & Architectural Leadership
Broadest Enabled Ecosystem
Breadth of solution for any class of Big Data Analytics platform
Intel Information Technology Intel Information Technology
Res
ponsi
veE
ne
rgy
Eff
icie
nt
Hig
hA
vaila
bilit
yS
ecu
re
Intel’s Foundational Technologies Offer Advanced Solutions for Big data Analytics
Ch
oic
e
Intel’s Big Data Building Blocks
Intelligent Storage1
Scale-out Storage1
Scale-up Storage1
Intel® SSD 710 series, DC S3700 (SATA)
Intel® SSD 910 series (PCIe)
Intel® Ethernet Controllers
Intel® Ethernet Adapters
Intel® Ethernet Switch Silicon
Intel® True Scale Fabric
Compute Network Storage
Intel® Data Platform
Intel® Expressway Service Gateway
Intel® Cache Acceleration Software
Intel’s Lustre
Intel® VT and Intel® TXT
Intel® AES-NI
Software & Technologies
Intel® Xeon® Product Family E3-E5-E7
Intel® Atom™
Intel® Xeon PhiTM
Intel® Quark™
Xeon-based storage systems are available in a wide range of configuration options from the industry’s leading storage vendors
25
Intel Information Technology
• Build a cost-effective, versatile Big Data platform. One Size does not fit all.
• Technology is important, but skill sets are essential.
• Ecosystem is maturing. Easier than ever to get started.
Summary
Big data analytics has led to big value across every sector
Intel Information Technology
IT @ Intel: Sharing Intel IT Best Practices with the World
27
Learn more about Intel IT’s initiatives at www.intel.com/IT
Or @ajayc47
CIO and IT Perspective
IT White Papers, Audio-Video Blogs
IT-to-IT Community