The Ghost is out of HPE’s The Machine
By Zafar Anjum January 11, 2017
- The Machine will power innovations across a vast spectrum of business and societal applications
- Memory-driven computing enables massive data sets, making the memory hierarchy obsolete
HPE claims that The Machine that it has built ushers us into the age of Memory-Driven Computing and will power innovations across a vast spectrum of business and societal applications, from Artificial Intelligence and Cognitive Computing to medical research, smart cities and transportation power.
To find out more about The Machine, we spoke to Kirk Bresniker (pic, above), Hewlett Packard Labs Chief Architect and HPE Fellow/VP. Here is a backstage look on what The Machine is and how it will be used.
How do you envision the future of computing? What do we have beyond cloud computing, data analytics, Artifical Intelligence (AI), Internet of Things (IoT), robotics, blockchains, Augmented Realty (AR) and Virtual Reality (VR)? What is the fundamental shift that is coming in the world of computing to radically shake it up?
If you asked Google, Microsoft or IBM, all three of them have put machine learning (which combines elements of or is enabled by cloud computing, data analytics and AI) at the heart of their product strategy. Both Sundar Pichai and Satya Nadela, the respective CEOs of Google and Microsoft, have acknowledged this in their talks. IBM has been proudly presenting its machine leaning offering, IBM Watson, for a while now.
Only Hewlett Packard Enterprise (HPE) among the big boys club was missing from this race. However, in the last few months, HPE has opened its cards and has unveiled a very unique product, The Machine, (still in prototype) that claims to usher us into a new era of computing — from processor-centric computing to memory-driven computing. It also inaugurates the era of photonics in computing (using light, instead of electricity, to transmit data).
According to Hewlett Packard Labs managing editor Curt Hopkins, “The Machine is a computing architecture so radically different than any which has come before that it will affect everything we do in the future. Hewlett Packard Labs has spent the last five years developing the memory-driven computing, photonics, and fabric that has gone into The Machine and which have made the impossible inevitable.”
HPE claims that The Machine will power innovations across a vast spectrum of business and societal applications, and unlock scientific, technological, economic and societal breakthroughs that cannot be addressed or even imaged with today’s technology.
To find out more about The Machine, we had an exclusive media sharing session with Kirk Bresniker, Hewlett Packard Labs Chief Architect and HPE Fellow/VP. The sharing session took place in September in Singapore where Bresniker, an HPE Fellow (a title awarded to the corporation’s most innovative thought leaders), joined us from California via phone. He provided us a backstage look on what The Machine is and how it would be used.
But first, why was it needed? Why was it invented?
The Genesis of The Machine
According to Bresniker, we are on the verge of a data explosion and in fact, 90% of data was uploaded in the last three years and the internet is only getting bigger: by 2020, one hundred billion connected devices will generate far more demand for computing than today's infrastructure can accommodate.
Picture this for a second: three times population growth, two devices per person, 2GB data transmission on average per device. Now, the problem is that computer architecture hasn’t changed much in the past sixty years and the amount of energy we waste computing this way is incomputable.
Bresniker said that the age of cheap hardware is over, and secondly, the compute is not keeping up, as data nearly doubles up every two years (2013-2020). As such, there was a critical need for a new memory fabric because of the magnitude of data, velocity of data, the inability of current machines to secure data, time-consuming data integration, the need for real-time insights, and the end of cheap hardware.
Hewlett Packard Enterprise (HPE) wants to change all that with The Machine. It is HPE’s solution to the data deluge we are currently facing, representing the first chapter of HPE’s research into Memory-Driven Computing systems, he said.
What makes The Machine a cutting-edge Innovative solution?
Why is The Machine radical? According to Bresniker, there are at least two innovations in this product. One, memory-driven computing enables massive data sets, making the memory hierarchy obsolete. Instead what we have is a ‘universal memory’, which is faster to access and has unforeseen capacities.
Two, the use of photonics destroys distance, so much so that 160 racks can behave as a single server. Photonics allows us to create unlimited memory, said Bresniker, which is quite revolutionary in itself.
In addition, The Machine customises the hardware to the workload which reduces cost, is less complex and uses less energy and occupies less space.
The Machine operates with hardware plus open software stack that makes it ultra-efficient with million-node management and Exabyte-scale algorithms. It helps in gathering insights through data and to top it up, security is built in from ‘silicon upwards’.
The Machine is secure in an unprecedented way, said Bresniker: it is ‘always safe, always recoverable, without sacrificing programming’. It has the ability to protect itself, even against completely unknown threats. With this product, data is always protected: ‘in use, in flight, and at rest’.
The Machine has a simplicity of its own, said Bresniker, because it has only a few data layers. It works with MDS (managed data structure) formats, which has local function calls and non-volatile memory.
In terms of performance, it is lightning fast. For example, for content-based image/video search, it is 1200 times faster than the conventional search machines (that use in-memory computing or simulated search).
Next page: The Machine in business
Author Name :
By commenting below, you agree to abide by our ground rules.