Sunday, April 6, 2014

Panel on "Is Analytics the Killer App for Big Data"

TiE Silicon Valley hosted a panel on "Is Analytics the 'Killer App' for Big Data" on Wednesday, April 2, 2014.

The panel was moderated by IBM's Piyush Malik. Piyush started the meeting by setting the stage and providing context. Enterprises have been using analytical technique and statistical modeling for quite some time. A typical path has been  Reporting -> Discovery -> Analysis -> Alerting -> Forecasting -> Simulation ->Modeling -> Optimization -> Stochastic Optimization -> Cognitive Analytics.

The democratization of Data (due to the "Big Data" phenomenon) has allowed the application of analytics to take this leap forward. In the 1990s the data was on premise and technologies like Terradata were used to provide Descriptive Analytics. The cost for this was a prohibitive $1m / TB. In the 2000s, companies like Neteeza (acquired by IBM in 2010), Greenplum, asterData (acquired by Terradata in 2011) and Informatica used massively parallel architectures to bring this cost down to $100k / TB. Today, using Hadoop and Cassandra stacks with NoSQL and Cloud platforms, companies like MapR and Datastax and able to deliver Predictive Analytics moving towards Cognitive Analytics for a mere $1000 / TB.

While answering the question, "Why BigData?", Piyush said, "1 in 2 business leaders do not have access to the data that they need".

I was reminded of a talk (in the late 90s) by Bill Perry that I had attended (while I was at Sun Microsystems). He was recounting his experience when he was Secretary of Defense. He said that often times, he was asked to make decisions when he did not have all the required data. Yet, the impact of his decision could mean life or death for the soldier. "You have to trust your gut," he said. "Don't always lean towards what will make you 'popular'".

Arif Janmohamed, Partner at Lightspeed Ventures spoke next. He spoke to LSVP's investments in MapR and Datastax affirming their commitment to BigData and the opportunity that the VC community saw in BigData.

Chris Jones, Director of Analytics at AAA was the next speaker. He spoke of his experiences as a consumer and producer of Big Data through his career that spanned Intuit, Adobe, Zynga and AAA.

Sai Devulapalli, Director of Analytics at Ericsson spoke about how they use BigData at Ericsson to make predictions about:

  • Customers - Behavior, Experience, Churn
  • Devices - Uptake, Performance, Failure
  • Networks - Performance, Failure
  • Campaigns - Uptake, Network Impact
  • Services - Uptake, Performance, Failure, Network Impact
Janet George, Managing Director of Accenture Technology Labs spoke next. Previously, Janet worked at eBay and on Hadoop while at Yahoo. She spoke about life events for Small and Medium Businesses and the importance of data in the stages of Expansion, Consolidation and Change in Ownership.

The panel provided an overarching viewpoint on the impact of BigData in enterprises today. However, I felt that some of the bigger questions were left unanswered. No one was willing to talk about solving "Grand Challenge" problems. How about using Big Data to:
  • Find a cure for Cancer
  • Predict the next Earthquake
  • Drug Discovery
If, at the end of the day, Enterprises are able to use "Big Data" to predict which brand of toothpaste I am likely to purchase next, we should be pooling our collective resources elsewhere.









1 comment:

Sujatha said...

Well put. I feel learning statistical analysis should be part of mandatory high school education so that the next generation could be trained in big data analysis to solve the big issues.