The shift from static data processing to real-time analytics is a transformation of enormous implications. We are at the onset of a remarkable time in the marriage of business and technology where electronics are not only faster and more reliable, but more energy-efficient and life-enhancing.
Clearly, the next wave of electronics innovation has arrived with the proliferation of real-time analytics, as I noted in my keynote at the JEDEC Server Forum last week. There are thousands of businesses jockeying to maximize its value to consumers, who today use more than 20 billion connected devices worldwide. Behind the scenes, servers in the data center and the cloud are taking on greater capacity and maintaining a high degree of scalability while processing data much faster, and needing less power to do so.
Analysis in Milliseconds
I like to think of real-time analytics as the foundation of a new approach where the time for critical information analysis is measured in minutes, seconds and milliseconds, rather than hours and days.
The focus is shifting from relatively small, static data pulls to a near-instantaneous flow of Exabytes of information. As the time invested in harvesting data decreases dramatically, the amount of data and its potential value will continue to increase exponentially.
On the horizon, Samsung sees many products, processes and services benefiting, including markedly improved electronic speech; deep learning advancements in medical screening, genomics, prediction modeling, and game theory; new strides in image and facial recognition; and a shift to the fast lane for autonomous driving. Advancements in Artificial Intelligence will speed the migration to real-time analytics, making for a more customized and better connected Internet of Things.
Memory – Critical Linchpin
None of this, however, would be happening to the degree that it is, without memory rising to the occasion. In fact, I would say that the Big Data Era would be inching forward rather than moving into hyper drive, if it wasn’t for tremendous improvements in memory technology, particularly as they relate to the infrastructure for real-time analytics and to enabling a more connected, user-focused world.
Advances in Memory Technology
Yes, mainstream memory is morphing, and it is doing so in three complementary ways. First, DRAM is being “Internet-ized” to deliver greater value with easier-to-scale, higher density solutions that consume less power (advanced DDR4 and soon DDR5). Second, the move to squeeze maximum value out of every memory module is picking up steam to the point that by 2018, 50% of large organizations will have adopted in-memory computing.
Thirdly, new memories will have a major impact. We see the huge potential for HBM2 in the realm of high performance computing, with its exceptionally high bandwidth, extreme parallelism and reduced latency. And there’s also NVDIMM-P – the most promising solution for closing the capacity gap within the memory hierarchy, using standardized DDR extensions and proven NAND technology.
The Leading Edge and the Future
Today’s memory arsenal has already begun to accommodate Big Data, but the pace of advancement will quicken and its efficiency will improve by orders of magnitude over the next few years. Samsung is leading the industry to answer the global call of real-time analytics not only with our 10nm-class DRAM, but also with our advanced TSV packaging, high density DRAM chips (currently at 8Gb) , scalability now to 128GB, high capacity LRDIMMs and RDIMMs, and exceptionally high bandwidth (currently up to 3200Mbps).
Watch how memory plays out its pivotal role. It will help usher in the Fourth Wave of Innovation in electronics – which will embrace Artificial Intelligence in mastering the challenges of the Big Data Era with greater accessibility, in addition to more efficiency, capacity and speed than previously anticipated.