markrubens - Fotolia
IBM hopes a new flavor of its private cloud platform will help customers cook up and serve AI and machine learning capabilities, both on premises and in public clouds.
IBM's Cloud Private for Data, revealed at its annual Think conference in March and formally made available this week, is designed as a foundation on which users can build multi-cloud strategies and introduce AI and machine learning capabilities. The platform also integrates with Red Hat's OpenShift container application platform to work with not just IBM Cloud, but also AWS and Google public clouds, which underscores IBM's support of open source platforms.
Most users who seek to modernize their data architectures want to bring public cloud features to their on-premises data, rather than collect data from multiple locations and ship it all to public clouds, said Rob Thomas, IBM's general manager of analytics.
To that end, IBM Cloud Private for Data containerizes dozens of its legacy applications and tools, such as Cognos, SPSS and DB2, that help manage, cleanse and connect data, to then interact with applications and data residing in public clouds. The company has also mixed in its hybrid cloud technology that connects on premises to public clouds.
"It is all about having just one cloud architecture for how to deliver applications and data across private and public clouds," Thomas said. "And from a data perspective, it helps users establish the building blocks for AI," he said.
Over the past several years, IBM's various cloud strategies have been criticized as too proprietary, too late to market and, upon arrival, out of step with corporate users' desires -- especially large enterprises outside of IBM's core user base. But this time, some said IBM has the right approach.
"In a sense, they are fighting for their life," said Judith Hurwitz, president and CEO of Hurwitz & Associates LLC, a research and consulting firm in Needham, Mass. "When you get to that point, you pull back from what you are doing and take another direction. So far, I like what they have done with this new cloud strategy."
Building bridges to the legacy world
Rob Thomasgeneral manager of analytics, IBM
IBM's goal to bridge traditional data marts, data warehouses and data lakes and next-generation AI and machine learning technologies in an orchestrated manner should appeal to big enterprises that store lots of data across multiple applications, Hurwitz said.
"IBM is hoping to take the best of what they do in the larger enterprise and match that up with Kubernetes, microservices and open APIs and bring them all together," she said.
IBM's approach with Cloud Private for Data uses very different technology pieces, but shares the same intent as Microsoft's Azure Stack, which figures to be a formidable competitor. Both are designed to make it easier for users who create workloads in the public cloud to also run those workloads on premises.
Some remain skeptical that IBM can compete with Microsoft's growing momentum in the cloud services business.
"With Windows Server and its management tools, this could prove to be a really tough market for IBM to gain traction in," said a solutions architect with a large technical services provider. "[Cloud Private for Data] will have to gain extraordinary momentum among its core base of mainframe and cloud users if they hope to compete."
Others remain cautious about the appeal of IBM's inside-out approach. Most users still don't want to upload all of their data to the cloud and prefer that their on-premises applications interact with the cloud only when appropriate. It's unclear how IBM will convince larger corporations to use this capability -- and others in IBM Cloud Private for Data -- to introduce their first AI and machine learning projects.
"IBM indicated the platform is where users could start to move toward implementing AI and high-end analytics projects," said Charles King, principal analyst of Pund-IT Research. "But there weren't many details about how the platform would support them."
At the Think conference in March, IBM demonstrated Cloud Private for Data on its own hardware, and it accessed only the IBM Cloud. But, this week, IBM added support for Red Hat's OpenShift, improved support for MongoDB and Enterprise Postgres databases, and improved its dashboard to provide a common look and feel across IBM and non-IBM products. It also integrated its Data Risk Manager to help companies deal with the impact of the European Union's General Data Protection Regulation, which went into effect May 25.
And a series of microservices enables users to create data catalogs from data from far-flung silos across and outside the enterprise, tagged to who last touched it, what source it is from and how it can be used.