HPE DISCOVER: Creating intelligent mobility using Memory-Driven Computing from edge to cloud to core

[Curt_Hopkins]

Kirk Bresniker

By Curt Hopkins, Managing Editor, Hewlett Packard Labs

While currently only 10 percent of enterprise data exists outside a data center, by 2022 the majority of that data will be in a gateway or edge device, says a 2017 report by research firm Gartner. Data is quickly becoming the most vital raw material for every enterprise, but it needs to be refined into insights and those insights acted upon in near real time. Given the demands of information lifecycles we will measure in petabytes and microseconds, how will we get around the unshoulderable burden of sending that data elsewhere to be processed?

According to Hewlett Packard Labs Chief Architect Kirk Bresniker in his HPE Discover session “Creating intelligent mobility using Memory-Driven Computing from edge to cloud to core,” compute will go out to where the data is. That process has already begun but when it is complete, the computing landscape will look far different than it does today.

For a while we will continue to see improvements in speed and computation from both traditional architecture and from innovations like photonics, which will allow us to transport data over a meter of high-quality line in five nanoseconds. But even there we will hit a speed limit, the speed of light. So central processing has an expiration date.

Aristotle said “All men have a desire to know.” What this means, according to Bresniker, is that we seek out information for high-quality decisions. As we incorporate more features into our computers based on our own cognitive capabilities, those systems will also inherit our voracious appetite for information.

“Intelligence needs to flow out to where the data is, to the edge, to gain insight from it and to take action upon it,” says Bresniker.

SWaP

In order to grow computational capability at the edge where the data and the opportunities are growing fastest, we will need to address some of the fundamental differences we will experience in moving from the secure, reliable conventional data center, says Bresniker. To address this, he suggests borrowing concepts from aerospace engineering such as optimizing for SWaP, or space, weight, and power.

“As we push computation away from the reliable, robust, and controllable power of the data center into a place where security and control are less established, where heat and humidity are not controllable, how much power will be available?” he asks, “How far can I push computation to the edge?”

When we get to where we can’t go further, where do we go from there?

Bresniker will explain Labs’ approach in this context, including the Path Forward exascale project, memory fabric, and Memory-Driven Computing, and explain how a unified architecture can turn a disjointed collection of transportation modes into intelligent mobility, the vital heart of smart city economics.

He will also review the big question of our age, data. Clients, says Bresniker, need to understand the business processes they have, their data lifecycles, and how to resolve the two with their current technical capabilities.

“No one is immune. Your question shouldn’t be ‘Is anyone working on bringing machine learning and real time analytics to turn data that we’re throwing or giving away into unbeatable advantage?’ Your question should be ‘Are they working for me or my competition?’”

Title: Creating intelligent mobility using Memory-Driven Computing from edge to cloud to core

Session ID: B4913

Date: Tuesday, June 19

Time: 4:30 – 5:30pm

Register for HPE Discover 2018 in Las Vegas, June 19-21

Photo by Rebecca Lewington