At Think 2019, IBM brings AI to the multicloud but confronts formidable challenges

Considering how far behind its public cloud is against Amazon Web Services Inc. and Microsoft Corp. in market share, the technology giant has wisely chosen to step up its focus on hybrid and multicloud management as its next best hope for deep differentiation.

At this stage in the development of the cloud economy, it’s not clear whether enterprises will adopt multicloud as a long-term architectural end-state, or simply as a way station on the road to reliance on strategic public cloud providers. Nevertheless, the multicloud option is becoming more prominent in enterprise information technology strategies, as we’ve seen recently in moves by VMware Inc., Cisco Systems Inc. and others in this arena.

In many ways, IBM is now the multicloud solution provider to beat. At its recent Think 2019 conference in San Francisco, the company launched new solutions for planning, migrating, integrating, securing and managing applications and workloads across any public or private cloud and any on-premises IT environment. Its chief announcements comprehensively address such emerging enterprise requirements as:

• Planning and design of multiclouds: IBM has launched a dedicated team of consultants to assist customers on multicloud projects. The new IBM Services for Cloud Strategy and Design advises clients on how to architect comprehensive multicloud strategies, ranging from design, migration, integration, road mapping and architectural services to navigating their journeys to cloud. The service leverages IBM’s Cloud Innovate method, IBM Cloud Garage approach and automated decision accelerators to support clients in multicloud application development, migration, modernization and management.
• Acquisition, orchestration and managing the multicloud: The company also launchedthe new IBM Services for Multicloud Management. This solution supports self-service acquisition and management of IT resources across multiple cloud providers, on-premises environments, private clouds, legacy infrastructure and container environments. It integrates with the ServiceNow Portal to enable purchasing, orchestration, configuration, monitoring, maintenance and cost governance of cloud services and solutions from multiple providers.
• Reuse of containerized integrations on the multicloud: IBM’s new Cloud Integration Platform provides a self-service environment for deployment of containerized data and application integrations throughout Kubernetes multiclouds. This new solution presents a unified self-service interface for fast creation and deployment of integrations for any business data and applications; enables containerization of integrations for decentralized deployment close to the source; leverages existing integration architecture in new containerized instances; enables integration specialists to write, test and secure code once, store it in the platform and reuse it; offers the flexibility to deploy integrations across any environment with a Kubernetes infrastructure; supports integration of new systems with applications and data from existing systems; facilitates reliable, quick transactions through asynchronous processing; creates a real-time stream processing application using event streams that examines transactions and decides which merit an alert based on notification preferences; and leverages API management and events streams to build an application that monitors transactions from multiple sources and triggers an email from the source app.
• Securing the multicloud: IBM announced Secure Service Container for IBM Cloud Private on Z, which helps users protect their containerized workloads from internal and external threats. It announced IBM Cloud Data Shield, which lets developers enhance their applications with no code change to help protect sensitive data within protected areas of execution. It announced a new release of its IBM Cloud Key Protect key management service, which now integrates with IBM Cloud for VMware Solutions, Databases, IBM Cloud Kubernetes Service and Virtual Servers. The newly announced beta of IBM Cloud Hyper Protect DBaaS protects against data breaches and data manipulations by leveraging IBM LinuxONE pervasive encryption and IBM Secure Service Container technology. The new experimental release of Hyper Protect Virtual Servers can instantiate LinuxONE VMs to build out their public cloud solutions with Hyper protection. And the newly announced IBM Cloud Hyper Protect Crypto Service, which is slated for GA in March, provides encryption key management with a dedicated cloud hardware security module that leverages FIPS 140-2 level 4-based technology.

In addition to these platform-level capabilities, IBM is also retooling its application portfolio to play well across multiclouds.  In this regard, IBM’s most noteworthy announcement at Think 2019 was the elimination of the lock-in of its core artificial intelligence solution family — Watson — to its own public cloud offering.

Under IBM’s Watson Anywhere strategy, AI workloads are portable across any cloud where customers’ data resides, including such environments as VMware Workspace ONE. Through integration with IBM Cloud Private for Data, customers are able to run Watson Assistant, Watson OpenScale, Watson Machine Learning and other Kubernetes-orchestrated and containerized IBM AI services on IBM Cloud and on other public, hybrid or multicloud environments. ICP for Data is IBM’s open, cloud-native information architecture for AI that comes with Watson Studio’s advanced data science, data engineering and application-building capabilities using Python, R and TensorFlow.

IBM is positioning Watson Anywhere as a strategic thrust that may tip the multicloud arena more deeply in its favor. But that’s a highly uncertain prospect. Unlike the early part of this decade, when IBM established clear mindshare leadership in the incipient AI market, it has been struggling in recent years against the onrush of AWS, Microsoft and Google — its principal public cloud rivals — into the AI market. Watson has fallen significantly behind in momentum in the AI industry.

This latest announcement that Watson apps and workloads can now run “anywhere” might gain it some adoption in some enterprises’ multicloud environments. However, unless IBM sweetens its Watson Anywhere strategy with substantially innovative AI capabilities that are suited to multicloud deployments, it won’t significantly dislodge AWS, Microsoft, Google and dozens of other solution providers that have flocked into the AI arena over the past decade.

What kinds of sweeteners might push Watson more deeply into multiclouds? I doubt that IBM’s forthcoming “business automation intelligence” for Watson will do the trick. This new capability may help Watson gain traction with CRM, ERP and other cloud application providers that need to embed data-driven intelligence.

But the market for AI-driven business automation is now very crowded, so it’s not clear whether IBM is offering anything substantially innovative. AI-driven business automation — whether it be embedded in chatbots, robotic process automation, IT automation management, recommendation engines or other apps — is now merely table stakes for combatants in the cloud AI wars.

And even if IBM entices developers with every Watson Anywhere value-added feature imaginable, it may be “too little, too late” for it to regain momentum. The AI industry ecosystem has become steadily more open, limiting any one company’s ability to lock in customers to any proprietary cloud, on-premise or other platform.

Widespread adoption of standard AI abstraction layers is enabling more developers to build in any language they wish and compile their work for optimized execution in any framework, hardware, cloud, or server platform. Central to this trend are higher-level AI application programming interfaces such as Keras, shared AI model representations such as ONNX and cross-platform AI-model compilers such as NNVM and TensorRT.

Nevertheless, IBM significantly improved its AI multicloud value proposition last week, which puts it on track for realizing significant bottom-line growth in coming years. Here are key takeaways from what Rob Thomas (pictured), IBM’s general manager of data and AI, had to say on theCUBE at the event:

• Watson AI everywhere in the multicloud: “There’s no AI without IA, meaning you need an Information Architecture to support what you want to do in AI. We start looking into that, our thesis became, so clients are buying into that idea, the problem is, their data is everywhere, on premise, private clouds, multiple public clouds. So our thesis became very simple, if we can bring AI to the data, it will make Watson the leading AI platform. So what we announced with Watson Anywhere, is you can now have it wherever your data is. Public, private, any public cloud, build the models, run them where you want, I think it’s going to be amazing.”
• Watson AI portability in the multicloud over Kubernetes: “The way that we enable multi-cloud and Watson Anywhere is through IBM Cloud Private for Data, that’s our data microservices architecture running on Kubernetes, that gives you the portability, so that it can run anywhere, because in addition to, I’d say AI ambitions, the other big client ambition is around how do we modernize to cloud native architecture, more composable services, so the combination gets delivered as part of this.”
• Tools for managing multivendor AI across the multicloud: “[AI is] a maker’s market…. There are a lot of people that want to make their own, and try their own. Most organizations are going to end up with hundreds of different tools, for building, for running, this is why we introduced Watson OpenScale at the end of last year, that’s how would you manage all of your AI environments, whether they come from IBM or not, because you’ve got an organization has to have this manageable, understandable, regardless of which tool they’re using. The biggest impact that we see is when we pick a customer problem that is widespread and the No. 1 [problem] right now is customer service. Every organization, regardless of industry wants to do a better job of serving clients, that’s why Watson Assistant is taking off.”
• IBM’s AI pipeline automation tooling: “We have a unique benefit in IBM, because we have IBM Research. One of our big announcements this week is what we call Auto AI, which is basically automating the process of feature engineering, algorithm selection, bringing that into Watson Studio and Watson Machine Learning. I am spending most of time figuring out how do I continue to bring great technology out of IBM Research and put in the hand of clients, through our products.”
• Watson out-of-the-box AI services for fast customer value: “We brought together the analytics and AI units in IBM into this new organization we call Data and AI. That’s a reflection of us seeing that as two sides of the same coin, we really couldn’t keep them separate. We’ve really simplified how we’re going to market with the Watson products. It’s about how you build, run, manage your AI, Watson Studio, Watson Machine Learning, Watson OpenScale, that’s for clients that want to build their own AI. For clients that want something out of the box, they want an application, we’ve got Watson Assistant for Customer Service, Watson Discovery, Watson Health Assets. So we’ve made it really easy to consume Watson, whether you want to build your own, or you want an application designed for the line of business. And then up and down the data stack, a bunch of different announcements. We’re bringing out big SQL on Cloudera as part of our evolving partnership with the new Cloudera Hortonworks entity. Virtual Data Pipeline is a partnership that we’ve built with Actifio. So we’re doing things on all layers of the ladder.”

Visit this page to view other theCUBE interviews at IBM Think 2019. And check out what my colleague Dave Vellante had to say about the announcements at IBM Think.