Follow Datanami:
November 16, 2012

HP Shapes Strategy Around Big Efficient Data

Ian Armas Foster

While important to today’s large enterprise environment, the cost of running big data is not insignificant. The energy requirements alone are enough to make some executives nervous.

HP hopes to facilitate a more sustainable big data analytics environment with the release of their ProLiant SL4500 Gen8 server. According to Bounds and Jimmy Daley, HP’s Director of Smart Storage, Industry Standard Servers and Software, the new server takes up half the space and cuts down energy consumption by 61% compared to previous generations. The result, according to Bounds, is a significant cost reduction. “Between capital and operating expense savings, we can save customers a million dollars.”

“When we talk to customers,” said HP’s Daniel Bounds, Senior Director of HyperScale Product Marketing told us, “one of their top line items is energy consumption. We’re taking close to half the energy out of the equation by deploying this.”

According to Bounds, as HP repositions itself in the big data market, this is just the first in a series of “large announcements” over the next several months. For example, Vertica, an HP analytics platform, announced a partnership yesterday with tech consulting company Capgemini in an effort to expand their analytics reach.

One specific aspect that HP is championing in both this release and the Vertica-Capgemini partnership is predictive maintenance. From a big data perspective, it is costly in several ways to reboot a system running a data-intensive calculation, from run-time to energy consumption to data requirement. However, the nature of big data analysis, especially on the petabyte scale on which the new HP server operates, ensures that failures will happen at some point. Being able to predict those failures in advance could increase the efficiency of fail-safe systems.

HP claims to have achieved that. “We were able to add features like predictive spare re-activation,” Daley said “where you can reboot time and rebuild drives four times faster if you have a predicted failure.”

That predictive maintenance capability feeds into the backbone of this Gen8 server: something Daley’s team developed called Smart Storage. Smart Storage looks to be a key to HP big data products for the foreseeable future as Daley explained what was involved in its creation. “From a storage perspective, my team has been focused on increasing the throughput in our controllers, keeping up with the adoption of solid state as that starts to get introduced into the enterprise market, and taking advantage of that with things like HP smart caching so you can buy some solid state and still maintain and keep the high capacity bulk storage behind it.”

The controllers are keys to the Smart Storage system, as they are able to keep the important data in the solid state, while keeping the rest in storage.

 “We’re getting a lot of performance out of the solid state devices by having our controller analyze the data and keep the hot data on the solid state while we keep the colder data out on the store,” said Daley.

Bounds also expects the server to integrate well with existing big data infrastructures such as Hadoop. Indeed, the server may have been designed with Hadoop in mind, as they ran their field tests on the open source big data analysis system to promising results. “All of the benchmarking that we’ve done on that platform have indicated that this is going to be a killer platform for Hadoop,” said Bounds. Of course, more will be known as the servers are actually sold and integrated. Regardless, Bounds sees HP becoming a player in the future of how people run their applications on Hadoop.

“Some of our customers are still running Hadoop as an offline activity to collect and analyze data and some of them are moving it into a more realtime analytics aspect,” Bounds said. “As those things evolve, requirements around latency and performance change.”

Again, this is just the first of many related big data announcements from HP as they move into the analytics arena. For now, their wish is to establish an architecture from which they can build up ensuing products and add-ons. “You’ll have other announcements from us; what we’re specifically talking about is a flexible converged infrastructure that creates a baseline for all of the exciting things that we at HP are going to do.”

Bounds is excited about HP’s big data direction. Should the rest of us be? Perhaps, as a big portion of the industry today is focused on reducing the myriad energy and data storage costs involved. If HP can deliver on that front, their big data strategy could prove fruitful.

Related Articles

Adding Autonomy to Personalized Medicine

HP Storage Lead Drills Down Trends

Microsoft Taps on Open Source Window

Datanami