In unveiling the latest versions of its Xeon server processors today, Intel amped up the excitement by announcing new monitoring capabilities made possible by new in-chip sensors.
Leading up to the Intel Developer Forum in San Francisco later this week, Diane Bryant, senior vice president and general manager of Intel’s data center group, talked today about new sensors that measure airflow, outlet temperature, utilization of CPU, memory, and I/O on the new Xeon E5-2600 v3 chips.
Not that the news captured the imagination of investors. Intel stock closed less than one percent above where it started for the day. Still, the positioning today points to some of the biggest areas of concern for Intel: the ability to provide ever more concrete data that companies can analyze to improve operations and find proof of energy efficiency.
Storage of analysis of data sits closer than ever to the center of Intel’s business. Data can scale out across more and more racks in data centers. Companies can employ architectures like the Hadoop open-source software, for storing, processing, and analyzing bulk data, or more cutting-edge systems like the Spark in-memory engine and its accompanying toolset. If big data weren’t crucial to Intel’s future, it would not have invested hundreds of millions of dollars into leading Hadoop seller Cloudera.
More sensors sniffing operational details on Intel chips can mean more data flowing into Hadoop file systems. And data analysts could very well find themselves exploring the data and telling their colleagues in IT about how much more efficient these new chips are, relative to previous ones. It’s sort of a cycle that could keep companies buying Intel chips.
Including more sensors on Intel chips could double as a solid demonstration of the value of sensor data, which Intel and other companies have been touting as a market opportunity. Intel has joined in on industry groups involving the Industrial Internet, or the Internet of Things, whereby devices connect with one another and can provide sensor-based, real-time data, which can then be used to adjust operations on the fly.
In essence then, today’s news could be interpreted as a new way to promote the Internet of things as a mere on-ramp for buying Intel chips.
And speaking of doing that, Intel executives today talked about Intel chips finding their way into more storage and networking devices, reflecting the continued rise of a computing style that software controls. VMware has led the charge on this approach, with its “software-defined data center” marketing term, implying that admins can decide how much computing, storage, and even networking capacity to allocate for certain types of applications. But Intel wants its hardware to provide the underlying infrastructure across these three layers of IT, not just in the server chips that deliver compute power.
And that’s a worthy diversification as ARM-based chip architectures get closer than ever to being ready for production use in companies’ data centers.
The strategy played well with analyst Patrick Moorhead.
“In the bigger picture, this announcement was about Intel solidifying their server lead but more importantly, giving people more reasons to move to Intel-based storage and networking,” Moorhead, of Moor Insights & Strategy, wrote in an email to VentureBeat. “If you buy into a software-defined world, you then need to pick a hardware platform. Intel wants it to be theirs.”
And in the domain of servers specifically, Moorhead walked away impressed that partners like Dell, HP, Lenovo, and IBM were able to create denser, lower-power, higher-performance boxes, he wrote.
That’s all good. But as AMD and others work up ARM-based chips that can catch attention because of their power consumption, sensors that can deliver granular information about resource usage and efficiency could represent a sort of secret weapon for Intel.