lundi 3 février 2014

Big Data goes big in 2014: Analytics for everyone

A lire sur:

Big Data analytics isn’t just greenfields and early adopters anymore. Vertica’s VP/GM says this year will mark the journey from ‘Big Data’ to no big deal.
Big Data is no longer the bogeyman that’s going to drown your enterprise—but most organizations are still carving out a path to powerful, predictive analytics. Discover Performance recently spoke to Colin Mahony, vice president and general manager of HP Vertica, who says that in the year ahead, more businesses will improve the return on their information—and the investment in exploiting it.
Colin Mahony
Colin Mahony

Q: Big Data has been on the enterprise radar for several years. What’s different about how IT leaders should be looking at Big Data in 2014?

Colin Mahony: I think 2014 is the year that massive shifts take place around Big Data. In 2013 there were lots of experimental and periphery Big Data projects, but I wouldn’t say there was widespread offloading or upgrading of existing systems to fully take advantage of today’s capabilities.
There have been a few early adopters, but now it’s shifting to the mainstream. People are looking not only at new applications of Big Data, but they want to replace their antiquated mainframe and other high-priced systems with next-generation information management architectures. They have information that’s been locked up for 30 years in expensive mainframes, and now they can start sharing that information. CIOs are realizing that beyond the hype of Big Data, there are incredible opportunities to improve and extend.
In Competing on Analytics, Tom Davenport wrote about three stages of analytics. The first phase he refers to as descriptive, and that’s where business intelligence has been for the last 20 years: You take some data, you create some pie charts and reports about that data, and it gets shared, mostly in printed material and on computer screens. Generally, only a small fraction of information today is illuminated even in this very basic stage.
The second phase, the one I think we’re in right now, he calls prescriptive. It’s about getting a lot more detailed data, but focusing in on the real opportunities and issues, and then taking action on them.
The third phase is predictive. Some of our more advanced customers have now moved on to that predictive phase, and that’s all about being able to tell the future before it happens. Prior to Vertica, many of them struggled so much with the first two phases, they couldn’t imagine what it would be like to operate predictively. We fundamentally change the game for so many customers.

Q: What are the characteristics of predictive business intelligence?

CM: Let’s take one of our customers as a great example—a large cable company customer had a hunch that they were overspending CAPX to expand their footprint into new cities. They believed it was imperative to maintain service levels at all times, and so, without any data to guide them, they overinvested across the board.
Our customer then did something very cutting edge, which was to load a massive volume of its network log data into Vertica. They wanted to do a prescriptive analysis to figure out whether or not and how they could reduce their expansion spending so that they were not overinvesting. They immediately identified some correlations that revealed their CAPX spending was grossly out of pace with their true expansion. In short, several areas required a fraction of the expanded plant to maintain the exact same service levels. They then took it a step further and found that the new models could predict exactly where they needed to be in the future, which not only reduced CAPX but better informed them as to which markets they really ought to be in. In an 18-month period, they saved $302 million in CAPX by doing this analysis.
While this example was certainly bleeding edge at the time, we are now seeing use cases and deployments like this quite broadly in the market. This is one of the reasons that I believe analytics will really go mainstream in 2014. Both business and IT understand that there are incredible opportunities to monetize data, whether that data is big or small. In fact, what I also like about this example is that, in the past, no one would have really ascribed such value to network data, but in fact the detailed history it could replay proved immensely valuable.

Q: Some companies are tackling data problems on a one-off basis, while others are embarking on a more end-to-end transformation. What’s the most successful roadmap?

CM: Both approaches are valid, but what I see a lot is customers who don’t want to bet the farm on an end-to-end approach initially. Most customers start with a very specific use case, which is smart, given the amount of broad hype surrounding Big Data.
Everybody in IT is under enormous pressure to deliver a great return on investment on every project, and also the other ROI—return on information.  Given that so much of IT’s budget is dedicated to existing projects, it is natural for them to want to start small with a test to ensure it is a fit. Then, once they see it works, they start becoming much more comfortable with an end-to-end approach. I would say 70 percent of our customers start relatively small, and before long, they start trying to find places to put Vertica because they like what it does so much.
Now that use cases are becoming better known, I think we’re going to start to see larger projects get under way, and requests for proposals for not only greenfields but also more transitional applications. There’s research out there right now that says, on average, analytics projects in enterprises are delivering a 10-to-1 ROI. We’ve actually seen much greater returns as well.
If you do things right, you can get massive benefits, so you can justify a large project, but I also think it makes a lot of sense to find that specific use case, and make sure you’re picking the right tool for the right job.

Q: It seems like that targeted approach leaves a lot of value on the table. Do you feel like people are losing sight of the tremendous scope of their data and how much they could potentially do with it if they had the right analytic system?

CM: I don’t think they ever had the scope. So many companies don’t know the value of their data or what they can do with that information. Part of our job is getting our customers and our prospective customers to talk to each other, because they just teach each other so much about the power of what you can do with analytics. I think people are starting to see the light, so to speak, but of course all organizations must move at their own pace.

Q: Gartner’s top IT trends for 2014 made no mention of Big Data. Is Big Data just business as usual? Or is Gartner missing something?

CM: Gartner’s pretty good about looking into the future—they called Big Data before everybody acknowledged it was there. Instead, it probably reflects the fact that, at some point, no matter what you do, no matter what applications you’re building, Big Data is just going to be part of it.
When you use your GPS system, you don’t want to see all the analytics that are going on in the GPS to find your way. You just want it to tell you where to go, and I think that’s really what the world needs when it comes to information and analytics. Is it data and analytics? Absolutely. Under the covers, it’s there, but do I need to call it out separately? Probably not.

Aucun commentaire:

Publier un commentaire