By Charles King, Pund-IT, Inc. March 28, 2018
Computing “eras” are a common concept among those who work in or focus on the IT industry. One progression starts with the mainframe era (late 1950s to the present) and the client/server era (1980s to present). However, the notion breaks down a bit in the third era which is variously called the digital, information and Internet era due to the inclusion of divergent ecommerce, cloud computing, mobility, analytics, IoT and other essentially web-enabled processes.
That confusion isn’t especially surprising since the first two eras are associated with specific systems or platforms that fundamentally altered the way enterprises and other organizations operated. In contrast, web-enabled compute processes tend to be squishier in terms of the platforms they require which continue to undergo massive evolutionary changes.
That touches on a key point of IBM’s THINK 2018 conference last week in Las Vegas. During her keynote, IBM CEO Ginni Rometty and guest commentators described how the company is working with global customers and partners to embrace a “new era of man + machine, an era of data and AI (artificial intelligence).”
That’s a fascinating idea, especially given the growing interest and investments in AI, advanced analytics and associated machine and deep learning processes. But are we really at the cusp of a new computing era or was Rometty engaging in mere cheerleading? Plus, if the former is true, how well positioned is IBM to provide its customers and partners the tools and technologies they need to succeed in the era of man + machine?
Something old + something new = Big Blue enterprise-class AI
To begin, what exactly does IBM bring to the table in terms of AI? That’s an interesting question since the company has been decidedly ahead of the curve and competition for so long. In fact, two senior IBM Research scientists (Claude Shannon and Nathan Rochester) along with MIT’s Marvin Minsky and John McCarthy organized the Dartmouth Conference in 1956 which many point to as the seminal beginning of AI development.
More recently, the 2011 winning performance of IBM’s Watson platform on the gameshow Jeopardy! put the company’s notion of cognitive computing front and center in the public consciousness. THINK 2018 saw IBM firmly shift from cognitive to AI as central to Watson’s brand—not a surprise given the market’s broad acceptance of AI as a defining term, along with consumer/home devices seemingly related to AI, including Amazon Alexa and Google Home.
How close to Watson are those products in terms of AI capabilities? Their makers argue that they’re running neck-and-neck with IBM, but the fact is that Amazon and Google barely occupy the same track. IBM has long focused on the problems and issues of primary concern to its large enterprise customers, and Watson’s capabilities and related services reflects that core strategy. Amazon and Google may talk-up the business-readiness of their products but the vast majority of their applications and use cases are clearly consumer-focused.
That doesn’t mean that consumer devices are entirely outside of Watson’s purview. One of the most intriguing announcements to come out of THINK was IBM and Apple’s extension of their strategic partnership with efforts to integrate and mobilize the pair’s respective AI capabilities. The new IBM Watson Services for Apple Core ML leverages IBM Watson for machine learning heavy lifting, then deploys resulting models on Apple’s iPhones and iPads.
As an example, the companies noted that Coca-Cola is running a pilot program on the service using IBM Watson to create machine learning models of the company’s soda dispensing machines. Maintenance personnel in the field are then using Apple iPhones to scan soda machines for signs of mechanical breakdowns, electrical problems and unauthorized alterations.
This may seem a bit mundane compared to asking Amazon Alexa to order a birthday present for your spouse or Beyoncé’s latest tracks. But when it comes to technology, businesses are mostly interested in: 1) saving money and/or time, 2) increasing commercial opportunities, and 3) reducing operational complexity. It’s likely that other vendors will eventually get off the sidelines when it comes to enterprise-focused, AI-based, mobile-enabled services, but for now, IBM and Apple are the only game in town.
The platform is the computer
Another notable issue at THINK 2018 was IBM’s rhetorical use of platforms to both define and propose man + machine solutions for enterprises’ problems. Some may see that as little more than a marketing gambit, but in practice it enabled the company to regularly trumpet the unique characteristics of its homegrown POWER9 processors, related Power Systems offerings and Z Systems mainframes.
The larger point was that for many use cases and many customers, solutions other than IBM’s are literally second-best. Power Systems and POWER9 both offer good examples of how this works to IBM’s advantage. IBM is promoting the former as a “modern data platform” that offers the foundation for numerous advanced analytics workloads, like the new IBM Cloud Private for Data solutions and services announced at THINK.
During a THINK breakout session entitled, “Preparing Your Data for AI,” Rob Thomas, GM of IBM Analytics, compared the information access, management, governance and integration processes required for effective analysis to the steps in shipping cargo. For decades, the shipping process was a time- and money-wasting snarl of inefficiencies. Then, in 1956, trucking entrepreneur Malcolm McLean invented standardized containers, an idea that fundamentally transformed the shipping industry.
McLean’s invention is analogous to containerized IT offerings, such as the Docker and Kubernetes open source platforms that are used to used to package software into standardized units for development, shipment, deployment and management. Leveraging the Power Systems-based Cloud Private servers introduced in November, IBM’s Cloud Private for Data is a “no assembly required” container-based solution that can be deployed internally as a private cloud or easily integrated with IBM Cloud.
Cloud Private for Data can enhance a wide range of information resources and analytics processes. Since it also supports IBM Watson Studio functions, including artificial intelligence training processes, Cloud Private for Data provides customers, as Thomas put it, “a ladder to AI.”
The POWER of AI
The singular characteristics of IBM’s POWER chip architecture were on full display at both THINK and the OpenPOWER Summit which ran concurrently with the conference. One of the Summit’s showstopper moments occurred during the opening keynote when IBM researchers announced that optimized systems with POWER9 CPUs and NVIDIA V100 GPUs achieved a breakthrough in AI performance.
Utilizing an online advertising dataset released by Criteo Labs with over 4 billion training examples, the POWER9/NVIDIA system trained a logistic regression classifier in just 91.5 seconds. The previous leading benchmark score (which used TensorFlow running on the Google Cloud platform) took 70 minutes to train the same model. In other words, the POWER9/NVIDIA system shattered the previous record score by 46X.
During the same keynote, representatives from Google (an OpenPOWER founder, along with IBM, NVIDIA, Mellanox and Tyan) described the company’s homegrown, POWER9-based “Zaius” two-socket systems. At past OpenPOWER Summits, Google was extremely tight-lipped about its efforts and plans around POWER9. So it was something of a surprise that the company’s representative announced that Zaius systems are supporting production workloads in the company’s data centers.
Interestingly, though POWER9’s computational capabilities make it a star for AI, its highly robust and flexible memory and networking capabilities make it an ideal platform for a wide range of intensive, high performance workloads and applications. In fact, another OpenPOWER Summit keynote speaker focused on those points during a presentation on the Oak Ridge National Laboratory’s Summit, an IBM POWER9/NVIDIA-powered system that is likely to be the world’s most powerful supercomputer when completed.
POWER9 and Power Systems weren’t the only IBM platforms highlighted at THINK 2018. In particular, Z Systems mainframe solutions’ encryption/security features and blockchain capabilities got a lot of love from both company representatives and enterprise customers offering their own takes and testimonials.
Time and again, the 30,000+ conference attendees were reminded that IBM’s ability to continually reinvent Z Systems is a prime reason that, unlike geologic eras that fossilize and fall into dust, the “mainframe era” remains vital today. That point bodes particularly well for still emerging, future-focused technologies, including the IBM Q quantum computing systems that were also highlighted at THINK 2018.
Bringing partners and customers on the journey
IBM PartnerWorld, a conference dedicated to the community of value-added resellers (VARs), cloud service providers (CSPs) and other channel companies selling and leveraging IBM technologies, also ran concurrently with THINK. Some folks don’t readily associate IBM with channel partners, mainly due to its traditional focus on enterprise customers that directly purchase its solutions and services.
But one representative of IBM’s Global Business Partner (GBP) organization pointed out that while the majority of IBM’s sales are made to some 14,000 large global enterprises, its channel partners engage with a critical constituency of over 10X that number of small to medium-sized enterprises (SMEs). Moreover, since those partners and their SME customers are facing the same technological, business and market issues disrupting larger enterprises, IBM is well-positioned to help them face and overcome those obstacles.
The process isn’t always easy. In a one-on-one meeting, Jacqueline Woods, CMO of IBM GBP, pointed out that there has been a paradigm shift in the purchasing habits of technology customers. The increasing commoditization of IT means that sales, delivery and service models are more important than ever. Those are processes for which IBM can act as a trusted advisor and facilitator.
Along with its core hardware and software products, the company offers numerous solutions partners can use to create incremental value-added offerings of their own. IBM Watson, for example, provides the means for analyzing customer data to discover and develop new routes to business and sales. Additionally, IBM Cloud provides access to a host of offerings, including support for hybrid- and multi-cloud implementations and enterprise-class security services that can help channel partners satisfy SME customers.
I came away from PartnerWorld thinking that while some may not associate IBM closely with the channel, the company itself is dedicating a great deal of thought, time, energy and investment to satisfying its partners and their customers and communities.
By their nature, IT vendors’ annual customer and partner conferences are self-promotional affairs. There’s no shame in that since most in the audience want to see and understand what’s up with the host vendor’s latest/greatest products and achievements. There’s also more than simple curiosity at work here since those same customers and partners want to know how their trusted vendor will help them to work more efficiently and profitably.
IBM addressed those points in full at THINK 2018 and the adjacent OpenPOWER and PartnerWorld conferences. The company highlighted its current crop of key hardware, and software solutions and services, as well as new and improved offerings that will arrive in short order. Just as importantly, it demonstrated sometimes jaw-dropping technologies that should and probably will change the way people work and live in years to come.
Most importantly, however, IBM drew logical lines that effectively linked together its past, present and future. That made CEO Ginni Rometty’s vision of “an era of man + machine, an era of data and AI” entirely sensible and achievable, and established IBM as a vendor that will help its customers and partners achieve those goals.
© 2018 Pund-IT, Inc. All rights reserved.