Oracle Showcases Sparc M7’s Breakthrough Capabilities To Accelerate Software, Secure Memory
Wednesday’s closing keynote at Oracle’s 2015 Open World conference had a uniquely hardware-focused bent, with the software giant showcasing the advanced capabilities of its Sparc M7 microprocessors that implement “software in silicon” through a series of specialized co-processors.
A day after Oracle founder and CTO Larry Ellison discussed one of the main features of the M7 — Silicon Secured Memory (SSM) — Oracle’s engineering leaders explained the chip’s many capabilities beyond security.
The new chips achieve a revolutionary leap by enabling in-memory algorithms to accelerate databases, isolate memory sectors, and accelerate compression, encryption and SQL performance. They also can help developers identify memory bugs.
The M7 — the sixth Sparc processor Oracle has released in the five years since its acquisition of Sun Micro systems — represents a historic technological breakthrough, said John Fowler, senior vice president of systems, during the keynote.
The chips, first announced at last year’s Open World, have been beta tested by 25 customers and are ready to ship inside Oracle’s converged systems.
The M7 is the fastest microprocessor in the world, according to Oracle.
Such “preposterous claims” are backed by empirical facts, Fowler said in a session with the media after the event. Oracle has set 20 records with the chip — some for specialized applications, others for common CPU benchmarks.
With its 32 internal co processors, the M7’s time has come with the advent of the cloud, Fowler said.
“As exciting as the cloud is for customers, it amplifies our technical issues,” he said. “We have the intellectual, emotional and economic imperative to find better ways to do computing.”
Development of the M7 consumed Oracle’s hardware team for five years, Fowler said.
The project started with the goal of driving database performance, he said, but the hardware team soon “realized this idea was way bigger than just databases.”
Juan Loaiza, senior vice president of Oracle systems technology, told Open World attendees during Wednesday’s keynote that Oracle’s focus is on advancing databases and software by pushing integration across the entire stack.
To achieve order-of-magnitude improvements powered by the M7, Oracle has been writing specialized software, including updating the Solaris operating system, Loaiza said.
The many M7 co-processors accelerate performance of several crucial database processes, but running databases in-memory also presented “potentially a giant step backward in terms of security,” Loaiza said.
It’s easier to corrupt or destroy data in-memory than on storage, he said.
But with a co-processor that implements Silicon Secured Memory, as discussed by Ellison in Tuesday’s keynote, Oracle solved the problem without regressing security capabilities.
Just the opposite — the M7 yields a level of security beyond anything on the market, according to Oracle, through a breakthrough technique of protecting data from software culprits by tracking memory allocation.
Embedding algorithmic functions in hardware, and allowing those accelerators to talk rapidly across the “inter-chip network,” is how Oracle will drive “continued step functions in capability” through the coming decade, Fowler said.
The M7 is already used in-house at Oracle to assist software development because its security features can identify memory bugs in addition to locking down memory from malware, he said.
Andy Canham, co-founder of Eclipsys Solutions, an Oracle-aligned systems integrator, told CRN the new chips represent more than simple Moore’s Law enhancements
“Oracle is beginning to put technology in silicon to address business challenges much further up the technology stack, like analytic acceleration and security,” he said.
Customers building private database cloud platforms that are shared within their organizations can leverage the accelerators for extreme processing advantages while still accessing best-of-breed solutions for generic workloads, he said.
As customers move to the cloud, they care more about service level performance and availability and less about what chip or underlying hardware is delivering it, he said.
Canham expects the SSM technology to pull Oracle well ahead of Intel and IBM as a preferred platform for Oracle software-based workloads.
Ravi Kumar, executive vice president of Infosys, a global Oracle partner, told CRN the chip is the culmination of the vision Larry Ellison expressed after buying Sun — to engineer a complete stack, from application to processor.
Using silicon to secure the cloud realizes that vision, and represents a huge differentiator for Oracle partners selling cloud technology, he said.
The M7 is the best attempt ever made to bring together hardware and software, Kumar said.
And because security implemented on a processor doesn’t impact performance, customers are going to be clamoring for it — they understand hackers just don’t know how to penetrate silicon like they can software and operating systems.
“It’ll be a catalyst for moving workloads to the cloud,” Kumar said.
Fowler said during the media session following his keynote that he believes the M7 “is a big step forward toward what your data center is going to look like in five years from now.”
People will be skeptical of some of Oracle’s claims, he expects, and the next few months will see many proof of concepts.
“This is a truly ambitious engineering effort,” Fowler said. It would have been easier to implement those functions on an external co-processor, but instead, Oracle had to “touch just about every part of the chip to do it.”
“This is engineering lunacy what we did here,” Fowler quipped.