1. 2nd Screen’s Impact on
Big Data
Debi Stack
Director, Industry Consulting
Telecom, Media, Entertainment, Sports & Hospitality
2. Big Data
Media & Entertainment
companies are trying to extract
value from mountains of
structured and unstructured
data to monetize content with
relevant, personalized ads
“Technology has always been about enabling human potential.” - Michael Dell
2 Confidential Data Center Solutions
3. Content Everywhere
From new geographic opportunities for content distribution, to broadening audience demographics
and reaching new potential subscribers, multi-screen delivery represents perhaps the most exciting
digital media growth opportunity since the early days of the industry – Frost & Sullivan
The convergence of cloud, mobile, social and big data is going to completely transform the world
we live in by bringing levels of automation and optimization beyond what we can comprehend with
our understanding of technology today. – GigaOm
The total interactive advertising market generated roughly $34 billion in revenue in 2011 and is
expected to more than double to $76 billion by 2016. Of that, mobile advertising is expected to
explode from $1.65 billion in 2011 to $8.24 billion in 2016. -Forrester
650% Data 2.5 80% Data
Growth Quintillion Growth in
over next 5 bytes of Video &
years data per day Images
Geo
Personalize
Blocking Optimize
Content by
based on IP QoE
Zip Code
Address
Data Center Solutions
4. We’re
listening
Large volumes of structured
and unstructured data
coming from a wide variety
of sources
Inability to run queries across
these different data types
4 Confidential Dell Data Center Solutions
5. Big Data enables personalized shopping experie
5 Confidential Data Center Solutions
6. How do companies use Big Data?
DELL.COM U.S. CABLE PROVIDER POLITICAL CAMPAIGN SMART GRID
TOP PERFORMING INTEGRATION OF CPNI TURN 12 TERABYTES OF 350 BILLION METER
RETAIL WEBSITE & EXTERNAL DATA DAILY TWEETS INTO READINGS PER YEAR
*COMPUWARE–BESTOFTHEWEB2011 SOURCES SENTIMENT ANALYSIS
SECURITY SOCIAL MEDIA
100’S OF LIVE VIDEO PROFILE CONSUMERS
FEEDS TO PREDICT TO IMPROVE
THEFT MARKETING CAMPAIGN
U.S. TELCO GLOBAL TELCOS
ANALYZE 500 MILLION CONNECT 50 BILLION
CALL DETAIL RECORDS IP DEVICES BY 2020
PER DAY
2+
FINANCE RETAIL MILLION ADVERTISING
MEET REGULATORY & 1 MILLION WALMART INCIDENTS 30 BILLION
COMPLIANCE TRANSACTIONS PER RESOLVED IN AD DECISIONS DAILY
REQUIREMENTS HOUR 2011
6 Confidential Data Center Solutions
6
7. 2nd Screen Sports
Live Stats
Twitter Feed
Player Data
Video Streaming
Animation
Graphics
Camera Angles
Replays
Behind the Scenes
Social Media Links
Enhanced Fan Experience
“Whatever is on schedule that day, if
cameras are on it, we’ll stream it” -
Rick Cordella, vice president and general
manager of NBC Sports Digital Media
7 Confidential Data Center Solutions
8. 2nd Screen
Advertising
Since no third-party software has yet
been designed to manage multi-
screen audiences, AT&T sought to
build its own private-cloud
infrastructure ‒ one that would
enhance the scope of information it
made available. This would allow
AT&T to use its television and mobile
subscriber data anonymously to reach
similar audiences across all three ad
mediums.
“The better we match an ad to a consumer’s
interest, the higher the response rate, which means
advertisers see a better return on their ad spend,” said
Joshua Koran, AVP Product Management, Research
and Data Monetization at AT&T
• 20 millisecond bidding process
• 700 million anonymous user profiles
• 30 billion ad decisions daily
• 1.5 million customer attributes
8 Confidential Data Center Solutions
10. We’re
listening
It’s complicated
It takes too much time
and energy to figure it
out and then deploy it
The word 'innovation' is a marketing mantra these days, nearly up there in overuse with the likes of 'cloud.' But
innovation doesn't always mean a successful business can emerge from a new invention; similarly, innovation
should be thought of as reinvention of business processes as well. – 451 Research
10 Confidential Dell Data Center Solutions
11. Apache Hadoop and “Dell … was one of the first of
the hardware vendors to grasp
Cloudera Solutions the fact that cloud is about
provisioning services, not
about the hardware.”
Maxwell Cooter, Cloud Pro
Proven solutions
Excels at complex big data analyses across
large collections of structured and Proven components
unstructured data
• Hadoop handles a variety of Cloudera® Apache Hadoop software
workloads, including search, log
Crowbar Software Framework with a Hadoop
processing, data barclamp
warehousing, recommendation systems
and video/image analysis PowerEdge C2100, C6100, C6220, R720, R720XD
• Empowers organizations to work on Force10 or PowerConnect switches
modern scale-out architectures using a
clean-sheet design data Reference Architecture
framework, without vendor lock-in
Deployment Guide
Joint Service and Support
Partner Ecosystem
11 Confidential Data Center Solutions
Big Data Solutions
12. More Efficient Analytic Ecosystem
One Platform, Many Use Case Scenarios
Analytic
Offload
High Performance “EDW”
Analytic Data Mart
Data Warehouse
Accelerator
Big
Analytics In-Memory
On-Demand
Servers Storage Services
12 Confidential Data Center Solutions
13. Yahoo – “Yahoo launches big data analytics tool
for online advertisers”
http://www.computerworld.com/s/article/9227146/Yahoo_launches_big_data_analytics_tool_for_online_advertisers
Business Challenges ParAccel-enabled Results
• Many different data types / sources • Key enabling technology to drive year-to-
• Large volumes of data made it difficult to year growth in revenue of 35%+
separate signal from the noise • More effective targeting of ads through
• Highly perishable data, instantly fine-tuned behavior-based segmentation
perishable impressions of online audience into 10,000 groups
• Mixed workloads at scale • End-user driven rollout of new analytic -
from request to delivery in under 10
minutes
Analytic Requirements
• Optimize online monthly media placement
• Scale analytics across 20,000 dimensions across 10B+ media impressions and 20K
and 1B+ events daily ads
• Support for 24/7 ad hoc analysis
• Need for detailed data – not samples (Acquired Dec 2011)
• Scalability – Each month they
process/analyze 450TBs of incoming
data
• 400M unique users with 20B+ user
observations
Confidential
13 Data Center Solutions
• 10B+ media impressions Copyright 2012 ParAccel, Inc.
14. Merkle:
Simplicity & Time to Process (Marketing)
Business Challenges ParAccel-enabled Results
• Growing Data sets – analytic eco-system • Reduced processing times from 3 weeks
not scaling to accommodate to < 3 hours
• 3 week process to prepare data • Support multiple clients with a single
• Multiple analytic environments to platform
accommodate specialized requirements • Met SLAs plus incremental “on demand”
and meet performance SLAs client requests
• Reduced Complexity and Administrative
Analytic Requirements requirements - 1000+ steps down to just 6
• Faster data load and query performance
• Easy Ecosystem integration Business
Intelligence
• Scale easily to accommodate 1000s of
users
14 Confidential Data Center Solutions
Copyright 2012 ParAccel, Inc.
15. Nielson:
Near Real Time Analytics (Ad Placement)
Business Challenges ParAccel-enabled Results
• Struggling to quickly turnaround actual • Provide customers with “real-time”
versus forecast performance metrics feedback on ad effectiveness across 50
• Delay in delivery of new analytic products performance metrics
• Needed an analytic platform that could • Increased revenue with real-time support
support continued expansion into new for custom analysis for individual
markets and channels companies
• Holistic view of ad and campaign
Analytic Requirements effectiveness across online and offline
media
• Retention of data detail • Expand support for on-going expansion
into overseas markets
• No aggregation into cubes
• Eco-system integration (Brio & Dell)
• Ability to analyze data across any number
of new data dimensions
15 Confidential Data Center Solutions
Copyright 2012 ParAccel, Inc.
16. Dell uncovers new insights and reduces IT costs by US$35 million with a
business intelligence solution designed for big data
Accelerated customer shipment time by 33 percent
Dell on Saved US$2 million by improving product quality
BI Integrated data silos
Reduced IT costs by US$35 million
Learn More ›
Increased agility
16 Confidential Dell Data Center Solutions
17. Dell MDC
is the result of
four generations of
modular data center
experience
17 Confidential Data Center Solutions
Dell Cloud Solutions
18. MDC Modules are Ready for
Rapid Deployment
8X reduction in cost “Basically, we’re a group of
engineers, and Dell’s
80% lower power consumption design philosophy made
sense from an engineering
standpoint. And after
PUE as low as 1.03 hearing what the Dell
engineers had to say about
5X the amount of compute density the way they would do
things,
what everyone else had to
Deploy in 30 days from site readiness
say sounded like
marketing.”
Marty Gauvin, Managing Director, Tier5,
Australia,
18 Confidential DataDell Cloud Solutions
Center Solutions
19. eBay drives down
transaction costs with
rack-n-roll and highly
scalable modular Company
data centers
eBay has more than 100 million
active users and drives $62 billion in
transactions —more than $2,000
Challenge every second
Keeping up with the demand is one of eBay’s biggest challenges.
People are on mobile devices looking for what they want, when they
want it, how they want to buy it, for the price they want to pay. “ By lowering our overall
Solution cost per megawatt, we are
ultimately lowering the
Dell Data Center Solutions (DCS) delivers racks of servers ready to quickly roll
into the data center (rack-n-roll). DCS also delivered Modular Data Centers to cost per transaction on
the roof of eBay’s Phoenix, AZ data center that use free-air cooling year- eBay. When facilities and
round and get a power usage effectiveness (PUE) measured IT come together, and are
at 1.043 according to Green Grid standards at 115°n the desert.
i
able to come up with that
Benefits ecosystem that balances
• Maximum compute power, for the lowest wattage possible, itself automatically and as
for the best TCO a result, our buyers and
• 1/2 the capital costs to build compared to previous data centers sellers win.”
• 4x the rack density, 1/2 the costs to operate Dean Nelson, Senior Director of
• Ability to deploy thousands of servers in hours, provision them in minutes Global Foundation Services, eBay
Read the Dell eBay case study
Watch the eBay case study video
See the Project Mercury Green Grid white paper
19 Confidential Data Center Solutions
20. What we are hearing from CIOs
Many organizations are unable to derive insightful value from
their data
• Held back by rigid data management systems
• Incapable of accommodating data volumes
• Unable to answer the questions they need to answer
• Unable to respond fast enough to the needs of the org.
• Legacy data systems can’t scale to keep pace
• Existing platforms are overwhelmed with too much data
• Costs are rising inn part due to vendor lock-in
Many organizations are left asking themselves, “How can
we get better at this? How can we more effectively and
efficiently store, process, analyze, and use all of the data we
are capturing?”
20 Confidential Dell Data Center Solutions
21. Take the Next Step
• Schedule time with a Next Generation
Computing Solutions Domain Specialist
• Learn more
PowerEdge C
– Dell.com/PowerEdgeC
Revolutionary Cloud Solutions
– Dell.com/OpenStack
– Dell.com/dcswa
Revolutionary Big Data Solutions
– Dell.com/Hadoop
– Dell.com/BI for Business Intelligence Services
Dell Modular Data Center
– Dell.com/MDC
21 Confidential Data Center Solutions
Good afternnoon. Thank y ou for joining us to talk about 2nd screen’s impact on Big Data
I can’t tell you how many times I’ve heard “I thought Dell was just a PC company”. AND Of course, Dell is prevalent in many of the world’s largest data centers & We have collaborated with some of the largest studios & TV stations, who have been very open with us about their requirements . As we look at 2nd screen we have products & services custom built for the M&E industry. For example, (Back to our PCs, we are deploying our new Precision workstations for Video rendering at major studios and Hadoop for Big Data Analytics with online search engines and advertising agencies along with Virutal Private Clouds to protect your IP and DRMNow that we are on the cusp of something new with 2nd screen, we will build exactly what you need and not what you don’t through our Dell Custom Solutions. #1 Cloud Infrastructure provider to OTT players. And, as Dell becomes more of a solutions company, with our acquisition of Perot & a well thought out M&A strategy to acquire cloud enabling technologies.we embrace Open Source community, bring in Open Stack Cloud & HadoopCloudera solution. TCO, Faster ROI, Deploy Open stack cloud in 2 hours instead of days. Saving real estate, save $1.6M per rack / per year based on power, density & shared infrastrucutre. Overflow in dcenter gave way to Modular Data center.
Big data growth, estimated at 650% over the next 5 years, is being created by not only human interaction but the growth of machine generated data. Ericsson predicts 50Billion IP enabled machines will be connected by 2020.Digital media, wireless technologies and the Internet have given marketers the ability to communicate their brand value to customers in ways not possible five years ago. Location based services, internet based advertising, mobile advertising, social media and the extensive use of “Big Data” analytical models have provided marketers new dynamic tools to identify their most profitable customers, speed up product innovation and tailor their messages to specific individual.For years, companies have heavily invested in technologies that not only allow them know who their customers are, but why they purchase the products they do. Location based services now add the where to that equation by allowing them to know, for example, where a specific customer is at 2:13pm in the afternoon on Tuesday December 23rd and what department they are visiting
With so many consumer touch – points (smartphones, tablets, internet, kiosks, gaming) The message must be “contextually’ relevant to the individual. We now have the ability to deliver product information, coupons or sale data to consumers at the right time and at the right place. Companies that are able to harness these tools effectively, and responsibly, have a significant competitive advantage over those that can’t. PRIVACY PII and Non personally identifiable info.Marketers now have a medium that pinpoints the current location of consumers and can predict where they will be on any day, at any time and on any occasion. In a recent article in the New York Times entitled “Government Aims to Build a ‘Data Eye in the Sky’”, Albert Laszlo Barabasi, a physicist at Northeastern University, said “There is a huge amount of predictive power in this (location) data. If I have hourly information about your location, with about 93 percent accuracy I can predict where you are going to be an hour or a day later.”Marketers no longer want to only to understand past behavior, but rather s to PREDICT FUTURE BEHAVIOR and then to change it. And now both the technology and the data are there to accomplish that .Resource and analytic rich firms such as Acxiom maintain a database of over 1,500 data elements covering 126 million households and 190 million individuals in the US. Combining this level of information with the predictive power of location based services creates marketing opportunities beyond what most consumers (or marketers) can imagine.Acxiom Corporation February, 2012 http://www.acxiom.com/Marketing-Services/ As a general rule, consumers are unaware of how much and what type of information can be gathered about them every day. It would take the average consumer more than 300 hours to read the privacy policy at the websites they visit each year, according to the high-end estimates of a 2008 study published in the technology policy journal I/S. What is on the horizon for privacy regulation is a new and dynamic construct, one that will remove much of the burden on consumers and place it back to the companies that serve them. HadoopProvides a distributed file system for data storageProvides the framework for running jobs written in MapReduce so you can run queries across a wide variety of data types
Elaine & her daughter Jennie go shopping: the scenario above is clearly possible given the information from her phone and a plethora of available PII known about Elaine collected from web sites, credit card companies, retailer loyalty programs, social networks, wireless carriers, handset manufacturers and many others (Personally Identifiable Information)Yahoo's Interclick unit, for example, relies on massive parallel processing of millions of clickstreams each day to spot suitable targets for advertising offers. The more recent and frequent the behavior and the higher the value of the transaction, the more likely you are to see a related advertisement. So if you shopped on a car site yesterday, don't be surprised to see car ads on sports sites, news portals, or other seemingly unrelated sites today.on-Personally Identifiable Information (Non-PII) includes data about Elaine’s mobile carrier, zip code, area code and Jeannie’s exact location as she walks through the mall Since Elaine is a frequent shopper and has a desirable demographic to certain retailers they will use companies like Demdex Inc. that helps websites build "behavioral data banks". These data banks tap sources including online-browsing records, online retail purchases and a database predicting a person's spot in a corporate hierarchy. Much of this is also performed by emerging companies called “Internet Scrapers” or “Screen Scrapers” such as Automation Anywhere. Automation Anywhere, http://www.automationanywhere.comThe real bet is the risk presented by data breaches and ultimate consumer backlash to the brands of major corporations who have not kept their side of the trust bargain with their customers. As consumers become more and more educated about how much of their life is known to people they have never met, and high profile data breaches continue to mount, there will be an ultimate consumer backlash to those companies that have not been transparent and proactive in protecting the trust of their brand with consumers. “Privacy by Design (PbD)” concept was Ontario Privacy Commissioner Ann Cavoukian that has served as a model for many organizations. Although a number of corporate privacy officers were practicing privacy by design before Cavoukian coined the term, she’s been the voice most responsible for formalizing the concept and advancing it with industry and fellow regulators. Cavoukian’s efforts caught the attention of Forbes, which in an article July of 2011, praised Intel, the Graduate Management Admissions Council and Location Labs as early adopters of the PbD approach. Dell has a number of specific offerings in the data security and Big Data market that can help companies move their data security requirements into the future. Kashmir Hill, July 28, 2011, “Why ‘Privacy by Design’ is the New Corporate Hotness”, http://www.forbes.com/sites/kashmirhill/2011/07/28/why-privacy-by-design-is-the-new-corporate-hotness/ As a general rule, consumers are unaware of how much and what type of information can be gathered about them every day. It would take the average consumer more than 300 hours to read the privacy policy at the websites they visit each year, according to the high-end estimates of a 2008 study published in the technology policy journal I/S. What is on the horizon for privacy regulation is a new and dynamic construct, one that will remove much of the burden on consumers and place it back to the companies that serve them.
Here are a few examples of how companies use big data today.Dell.com and Social Media Command center use Big Data Analytics to predict consumer behavior, encourage crowd sourcing and address customer service issues before they “go viral”Global Telcos – ericsson predicts 50B devices connected by 2020 (from refrigerator ‘out of milk” to grocery store aisle reminder that you need milkAd placement agencies processing 30B ad decisions daily
We look forward to Olympics 2nd screen experience. The 2nd screen can be used in the living room while watching the games or in the stadium for a unique experience that brings more fans to the game. I understand that the NFL”s biggest competitor is a 70” screen TV so teams are planning for 2nd screen experiences that provide more fan engagement and a personalized experience based on integrated data from CRM, POS & Ticketing systems.
Adworks selected Dell’s Hadoop & Par Accel solution to reach consumers across TV, Mobile & tablet applications. Resulting in higher returns on ad spend. The processing power needed to enable 20ms bid process across 700m user profiles is supported by Hadoop, using Open Stack and Par Accel for High Performance Analytic Data mart and Data Warehouse offload.
Big data spans three dimensions and produces multiple value propositions• Volume: Lots of data, lots of uses…Enterprises are awash with ever-growing data of all types, easily amassing terabytes—even petabytes—of information. – Turn 12 terabytes of Tweets created daily into an sentiment analysis of political messaging – Convert 350 billion meter readings per annum to construct smart grid architecturesVariety: Big data is any type of data…Internally gathered data such as text, sensor data, audio, video, and click streams, combined with external data sources such as Experian and Acxiom. Profound insights are found when combining and analyzing these data types together.– Use 100’s of live video feeds from surveillance cameras to predict theft – Take advantage of the 80% data growth in images, video and documents to improve marketing campaigns for new consumer products• Velocity: The beauty and value of immediacy... Time-sensitive processes such as catching fraud, Big Data must is most valued when used in real-time decision making.– Its not about understanding buying behavior, the holy grail is to change it – in real time– Analyze 500 million call detail records per day, in real-time, to predict individual customer churn likelihoodVALUE – the 4th C
The Dell Apache Hadoop SolutionStreamlines/simplifies deployment, reducing time to multi-node production from days to hours (6-30 nodes in one hour)Result: You can leverage all of your data to make better business decisions
Value proposition: Hadoop excels at supporting complex analyses, including detailed, special-purpose computation, across large collections of structured and unstructured data. The scalability of free, open source Hadoop running on industry-standard, purpose-built Dell PowerEdge C servers allows organizations to quickly pull together and analyze more types of data than ever before at a lower TCO than proprietary solutions. Dell’s Crowbar software can help you get your Hadoop cluster up and running 10x faster than comparable solutions. Validated designs, reference architectures and deployment guides can light the way, with Cloudera and Dell there when and how you need/want them.Key messages:Faster time to production: The Dell | ClouderaHadoop Solution streamlines/eases deployment, reducing time to multi-node production from days to hours. (6-30 nodes in one hour)Efficiency: Hadoop can be deployed in 1/10th the amount of time of comparable solutions and it’s open source. The Dell PowerEdge C2100 is designed to lower your total cost of ownership, saving on space, energy and costs.Low risk: This solution is built on Cloudera’s distribution of Hadoop, one of the leading distributions of Hadoop in commercial and non-commercial environments, with Dell’s performance-based PowerEdge C hardware in a known, tested configuration on proven architecture.Top to bottom support: Call Dell, we can handle it all, or warm transfer you to Cloudera. Because Hadoop is open source, community support is also widely available.
An initial starting point is as a high performance analytic data mart - e market basket analysis or fraud detection. In this case, if this is an existing application, the customer is often looking to replace the underlying database with ParAccel. In many cases this may be SQL Server, especially where it’s grown to be between 500Gb and 1 TB in size for analytic workloads. The customers in this case have already made the decision to move off of their existing database vendor and are evaluating other alternativesMany customers have already made investments (often tens of millions of dollars) in their existing EDW (Enterprise Data Warehouse) – often Teradata and/or Oracle – so Analytic Offload would be a good way to introduce Dell/ParAccel into these kinds of environmentsIn this case the Dell/ParAccel solution (Analytic Offload) would sit alongside the existing EDW – with data being copied in whole/part from the EDW into the Dell/ParAccel implementation.This enables customers to run their analytic workloads outside of the EDW where they were bogging down the performance of the entire EDW which is much better optimized for general reports / dashboardsAt a major Top 10 international retailer, this approach provided 180X better price/performance compared to the customer’s existing Teradata EDW. In that instance, the customer saved almost 50% on their upgrade and used a portion of the savings on a complete ParAccel solution that was able to do deeper and richer analytics than anything they had tried before. Summary:ParAccel on Dell – Deployment Options: - High Performance Analytic Data Mart for Speed of thought analysis typically around one subject area. - Data Warehouse Offload from your EDW (Enterprise Data Warehouse) for your most sophisticated ad hoc queries and data scientists - Analytics in the Cloud - Or – an all In-Memory Analytic deployment for near Real time analytic applications.Explanation:ParAccel on Dell can be deployed in a variety of ways depending on what’s the best path of action for the customer. An initial starting point is as a high performance analytic data mart - this is often related to a functional or departmental application like market basket analysis or fraud detection. In this case, if this is an existing application, the customer is often looking to replace the underlying database with ParAccel. In many cases this may be SQL Server, especially where it’s grown to be between 500Gb and 1 TB in size for analytic workloads. The customers in this case have already made the decision to move off of their existing database vendor and are evaluating other alternativesMany customers have already made investments (often tens of millions of dollars) in their existing EDW (Enterprise Data Warehouse) – often Teradata and/or Oracle – so Analytic Offload would be a good way to introduce Dell/ParAccel into these kinds of environmentsIn this case the Dell/ParAccel solution (Analytic Offload) would sit alongside the existing EDW – with data being copied in whole/part from the EDW into the Dell/ParAccel implementation.This enables customers to run their analytic workloads outside of the EDW where they were bogging down the performance of the entire EDW which is much better optimized for general reports / dashboardsAt a major Top 10 international retailer, this approach provided 180X better price/performance compared to the customer’s existing Teradata EDW. In that instance, the customer saved almost 50% on their upgrade and used a portion of the savings on a complete ParAccel solution that was able to do deeper and richer analytics than anything they had tried before. Additional areas where ParAccel/Dell may be considered is with enterprises or SaaS providers who want to deliver analytics via the cloud. The ability to do modular expansion provides re-assurance that companies can keep up with demand as it builds, without the need for a forklift upgrade at some future point in time. Since ParAccel is software-only, upgrades and expansions can take advantage of the latest and greatest servers and storage available within the Dell portfolioIn a more future-forward direction, future developments will be focused on how to deliver analytics “on demand”This could be used to provision new “test/dev” analytic environments as needed, or respond to short-time spikes in demand (example: quarterly analysis).
Summary – Yahoo! (Interclick data driven advertising) has built the leading data valuation platform for advertisers. This application, powered by ParAccel and running all “in Memory” on a Dell Cluster, pinpoints the most responsive digital audience amid the mass of online data and maximizes advertisers impressions and advertising investments.Interclick’s proprietary data valuation platform, Open Segment Manager (OSM). Our goal was to develop the most relevant and capable data valuation platform in the marketplace. Open Segment Manager (OSM) powered by ParAccelDelivers actionable consumer insights, scalable audiences and the most effective campaign execution. Aggregates and organizes billions of data points from third-party providers Analyzes vast amounts of diverse data across a myriad of sources Determines data’s value based on advertiser specific objectives to construct the most responsive digital audiences. Customized statistical treatments to provide insights that result in the greatest impact for brand marketers.Dell ConfigurationAll “in-memory” Dell cluster2 Dell R510 as leader/standby node32 Dell C6100 compute nodes
Merkle is a leader in providing individual and household information for leading retailers like Walmart.com, Target, etc. Merkle uses ParAccel, running on a Dell Cluster, to maximize the hit rate of their subscribers marketing campaigns by providing daily updates of household lists vs. monthly updates previously available. By reducing the processing time from 3 weeks to 3 hours, and the Data Management complexity from 1000+ steps to 6 steps, they are able to dramatically affect the response rate of their subscribers saving them millions of dollars and returning more value for their investment.Explanation:They are essentially in the business of providing the most effective and accurate information to these retailers for use in their online and offline marketing campaigns.It’s important to remember here that this is a business in which even single percentage point difference in accuracy/response can improve marketing ROI by over 100% Using their previous vendor, Merkle collected data from hundreds (if not thousands) of sources and spent more than 3 weeks cranking through the data to arrive at the most up-to-date view of each individual and household. This meant that they were unable to grow their revenue – short-term requests from customers were turned away because Merkle could not provide the data fast enough. If their customers wanted to slice and dice the customer set in new and unexpected ways, Merkle didn’t have a way to provide that view quickly or easily. Merkle also was unable to provide the most current view into their customer set with their 3 week lag – so they could potentially miss out on providing their customers with info on recent births or weddings - major life events that retailers will want to know about in order to make customized and relevant offers to their target prospects. A huge problem with their previous systems was that the performance limitations forced them to implement new clusters for each client, driving up implementation and maintenance costs (in a business where margins are pretty thin to begin with) With ParAccel, Merkle was able to fundamentally improve the way they do business with the ability to shrink their 3 week processing job to under 3 hoursThis enables them to keep on top of life changing events (like births and weddings) that would be of particular value to their retail customers They are now also able to handle short-term requests – boosting revenue by being able to generate an accurate ad hoc list for their customers. So for instance if walmart.com wanted to roll out a spanish-language diaper promotion to Hispanic mothers between the ages of 24 and 34 who have given birth in the last month, that is something that Merkle would now be in a position to provide.Merkle has implemented a single, scalable modular implementation that can accommodate all of their concurrent client requests from one system rather than many individual/dedicated implementations – this had further reduced the ongoing IT/DBA maintenance costs compared to their previous implementationMerkle have also greatly simplified their “manufacturing” process for producing the latest info by reducing the number of processing steps from over 1000 to just 6, requiring far less manual review and intervention
Summary: Nielsen is a well-known information service provider to numerous media companies - mainly TV and radio, but rapidly expanding to online.ParAccel on Dell enabled Nielsen to:Build new analytic expressions on the flyScale to meet expanding data requirements , provide real-time custom analytics, and drive revenue from new customized analytic offerings.Speed up offline expansion to new markets, and expand online web-based media consumption toward younger demographic audiences. Explanation:They were struggling with the performance of their previous platform and realized that they needed to upgrade to an analytic platformWithout it, they would have struggled to expand their international presence in new markets like China, Brazil and IndiaAs viewership grows in terms of online media (Hulu, YouTube, etc.) and the use of DVRs continue to rise, the ability to track ads and campaigns across multiple venues continued to increase in complexity making it more difficult to get a holistic view of viewer and listener behavior.Getting a good sense of actual versus forecast performance for shows, broadcasts and ads was delayed under the previous system, delaying Nielsen’s ability to turnaround information that their customers could use to make changes in placement and promotions. They needed a new analytic platform that could scale to meet their needsThey needed to be able to retain and analyze increasingly detailed online and offline behavior Building OLAP cubes would have taken too long and would have restricted analyses to a limited set of drill-downs The ability to rapidly analyze data along any dimension was key since they didn’t necessarily know what new data and dimensions would be available in the coming yearsThe new solution needed to be able to work with their BI/visualization vendor (Brio) and leverage Dell servers and storageWith ParAccel/Dell, Nielsen got a best-of-breed analytic platform that could track ad effectiveness across 50 performance metrics. They now have the ability to rapidly build and re-build new metrics for customer and internal use as needed As their data scales to many billions of records, Nielsen was able to implement real-time support for custom analytics – enabling their analysts to drive revenue with tailored analytics for individual clientsDell/ParAccel will enable them to continue their offline expansion to new markets and expand online as well to cover web-based media consumption, which is especially prevalent among younger demographic audiences.
Dell on Dell case study using Big data analysis to reduce It costs by $35M and accelerate shipment times by 33%
Power usage effectiveness – Green metric2010 AustraliaToday our PUE on EBAY is 1.018 measured in the desert at 115%
Rack, Stack & Roll. Needed more space. Up on the roof 4 MDC
The Dell Apache Hadoop SolutionBuilt on Cloudera’s distribution of Apache Hadoop, one of the leading distributions of Hadoop in commercial and non-commercial environmentsDesigned to work with Dell’s PowerEdge C hardware in a known, tested configuration on proven architecture. The solution also comes with top to bottom supportResult: You don’t have to take the risk
Non-Personally Identifiable Information (Non-PII) includes data about Elaine’s mobile carrier, zip code, area code and Jeannie’s exact location as she walks through the mall Use this slide for externally facing presentations
HadoopExcels at supporting complex analysesScalable, free, open source: Runs on industry-standard,purpose-built Dell PowerEdge C serversEnables quick data analysis at a lower TCO