Intel's Big Data Play: 5 Facts

Intel ruled the PC era. Can it do the same in the age of big data and the Internet of Things?

Michael Endler, Associate Editor, InformationWeek.com

February 20, 2014

6 Min Read

Last month at CES, Intel demoed tiny, low-power chips to power the coming wave of wearable tech and other smart objects in the Internet of Things (IoT). This week, the chipmaker introduced Xeon processors aimed at the other end of the spectrum -- high-end datacenters, where IoT and other analytics-intensive trends have increased demand for a new breed of hardware optimized for big-data applications.

Intel claims its new Xeon E7 v2 family boasts plenty of muscle to chew through numbers, with twice the average performance and four times the I/O bandwidth of previous generations. But is the chipmaker prepared to rule the top-of-the-line datacenters of the future? Here are five things to consider.

1. The new Xeons shine a spotlight on in-memory analytics.
"I don't know if this product line will be the tipping point for in-memory database adoption... but if it's not, then the tipping point is very near," IDC analyst Kuba Stolarski told us in an email.

[Has Windows 8.1 won over Windows 8 critics? Read Microsoft Sells 200M Win 8 Licenses: Yawn.]

Indeed, research firm Gartner projected last year that by 2015, 35% of midsized-to-large enterprises will adopt in-memory analytics, which speeds up analysis by placing a dataset in the system memory rather than in traditional disks. Intel claims the E7 v2 family boasts triple the memory capacity of previous generations. The chip giant this week touted that eBay achieved roughly twice the performance of current baselines during a recent trial with E7 v2 chips and SAP's Hana in-memory analytics software.

2. Intel's chips could take share from RISC systems.
Though Intel's x86 chips dominate the PC and server markets, many legacy mission-critical systems rely on RISC (reduced instruction set computing) architectures, which are engineered to provide improved performance by sacrificing instruction complexity for a much faster execution rate. Chips from companies such as IBM and Oracle usually power such systems, but if Intel's E7 chips can displace RISC in the next generation of high-end hardware, the company could command elite status in a world of increasingly data-driven, intelligence-hungry datacenters.

Figure 1: Xeon Processor E7 v2 product die (Source: Intel) Xeon Processor E7 v2 product die
(Source: Intel)

Intel claims E7 v2 can deliver up to 80% better performance and up to 80% lower total cost of ownership compared to an alternative RISC architecture based around IBM's Power 7+. The new silicon also boasts RAS (reliability, availability, and serviceability) features that originally appeared in its Itanium products -- the Linux-oriented family of mission-critical processors, mainly used by HP, on whose space the Xeons are now encroaching. The company also possesses peerless manufacturing resources, which should help it to produce new chips at yields and margins few competitors can match.

"Intel is upping its game," Gartner research director Sergis Mushell said in a phone interview. "They make larger chips that no one else can make."

3. E7 v2 could give Intel a needed win in a lucrative enterprise market.
In January, Intel reported strong demand for chips to power cloud servers, but sales for other datacenter products missed analyst expectations. The new E7 v2 chips could change that. They're very expensive (the top-end version is almost $7,000), but they're also aimed at customers who don't cut corners.

"Pricing of chips in the consumer market is going down, but as you look at the high-performance market, companies pay for the highest quality products at the highest speeds," Mushell said.

"The top-tier service providers have traditionally gone for top-bin parts," Stolarski said. He said businesses might not go for Intel's high-end

Next Page

32-socket configurations "due to price/performance inelasticity as you climb above 2-socket [options], but "they will still likely purchase the highest-end chips with a maximum number of cores."

4. Intel continues to invest for the Internet of Things.
Intel has shown a strong interest recently in IoT. From Intel Developer Forum to CES, most of the attention has gone to small, low-energy chips to be embedded in smart objects. These processors don't do a lot of number-crunching locally, so as IoT applications grow more sophisticated, more and more of them will rely on analytics work done in the cloud. With the new E7 chips potentially providing this back-end computational force, Intel is positioning its chips across the entire IoT spectrum.

"IoT, if nothing else, is the massive proliferation of data gathering points across an exploding array of devices and machine sensors," said Stolarski, noting that many of that data's uses, such as fraud detection and prevention by credit card companies, demand that complex analytics be performed, not only on the fly, but also "in seconds."

"Traditional databases on industry-standard servers just can't do that kind of processing. If Intel has found the right combination of performance metrics for its chipsets, it may set up the industry to finally realize the potential of the Internet of Things."

5. Intel still faces questions.
The E7 launched this week with strong industry support from Intel's manufacturing partners and generally optimistic notes from commentators. But Intel still faces challenges.

Stolarski said Intel's performance claims "indeed raise the Xeon family closer to some of the RISC-based mission-critical architectures." But he cautioned that RISC "will likely stay ahead in performance for the time being, with refreshes of their own coming up."

It "still remains to be seen if customers will migrate their mission-critical workloads en masse from RISC to x86," he said.

And, for all this talk about datacenter progress, don't forget about mobile, where Intel's chips are far outnumbered by ARM-based designs, said Mushell, who added, "Intel is always late, but this time, they were way too late."

He said the company has made efforts to realign its workforce, but that its mobile missteps speak to bigger issues. "Intel has [only] learned to lead," he said, noting that the company must learn to be more responsive to customer desires.

"They are changing their DNA, but DNA doesn't change that fast."

Engage with Oracle president Mark Hurd, NFL CIO Michelle McKenna-Doyle, General Motors CIO Randy Mott, Box founder Aaron Levie, UPMC CIO Dan Drawbaugh, GE Power CIO Jim Fowler, and other leaders of the Digital Business movement at the InformationWeek Conference and Elite 100 Awards Ceremony, to be held in conjunction with Interop in Las Vegas, March 31 to April 1, 2014. See the full agenda here.

About the Author

Michael Endler

Associate Editor, InformationWeek.com

Michael Endler joined InformationWeek as an associate editor in 2012. He previously worked in talent representation in the entertainment industry, as a freelance copywriter and photojournalist, and as a teacher. Michael earned a BA in English from Stanford University in 2005 and, pending the completion of a long-gestating thesis, will hold an MA in Cinema Studies from San Francisco State.

Never Miss a Beat: Get a snapshot of the issues affecting the IT industry straight to your inbox.

You May Also Like


More Insights