SANTA CLARA, Calif. — The future of cloud, according to John Considine, GM of IBM Cloud Infrastructure, is “solutions.” He’s talking about industry-specific products and services, and things like embedded artificial intelligence (AI), that make all that data sitting in the cloud more valuable to companies.
“All cloud providers are going toward solutions,” he said in the opening keynote at Cloud Expo 2017 today. “Why? That’s where all the value is. The other thing is: enterprises need them. Solutions are really the way to do end-to-end, solutions [are] where it’s at.”
Another trend Considine expects to see in the future is pooling resources in the cloud. “This disaggregation is very powerful to be able to combine those resources in new ways to deliver on an efficient infrastructure,” he said.
SDxCentral senior editor Jessica Lyons Hardcastle caught up with Considine at Cloud Expo to talk about the future of cloud. The following interview has been lightly edited for clarity.
Is hybrid cloud — and by that I mean on premises private data centers and public cloud — is that just a stop on the way to all public cloud?
Considine: I am a public cloud advocate. I do love the model and I think it’s the right thing. Once upon a time I thought that hybrid cloud would be a temporary thing. But after working on a lot of these solutions — with IBM, with our customers, with our partners — I think the hybrid cloud is going to be very durable and long lasting. The blending of your environments — things that can’t move, things you don’t want to move — necessitates hybrid cloud computing as far as the eye can see.
During the keynote, you said: “Solutions are the key and the future.” If I’m a company, what does that mean for me?
Considine: The idea of consuming cloud services is great. But it’s actually only a part of delivering solutions, or applications, or outcomes to your users as part of your business. You have to look at this bigger picture. Even using cloud, you still have your DevOps, your operations, how do you support it, how do you evolve it, how do you integrate it, what happens with the ecosystem. There’s a huge set of work around delivering value. And the cloud is only one piece.
How an enterprise should think about it is: they [solutions] go with the technology, with the ecosystem and partners, and then [ask] what is the full solution. The professional services, the help that we, and folks like us, can provide to build those solutions is really critical. It’s about learning what’s possible. Look at the emergence of AI [artificial intelligence], deep learning, machine learning — how do you know how it can be used? Now I can have packaged-up capabilities that expose these.
What role does Watson play?
Considine: Watson is an amazing, enabling technology. Think about artificial intelligence. We talk about conversation — deep learning, machine learning, all these different layers of it. Watson really brings this AI capability to a consumable point. AI is now a set of APIs you can call, and anyone can use that capability to enhance their business. We train Watson on medical, we train Watson on finance, we train Watson on contracts, we train Watson on customer support. So now Watson is this capability that can be blended with your business. That, combined with an enterprise data set, allows a set of new inferences and new learning, so then it becomes specific. It’s not contracts and contract language. It’s your business contrasts. Watson can become a part of the solutions, and you don’t have to gain all of the expertise on how to build it and how to advance it, it just becomes consumable.
Let’s talk more about creating resource pools in the cloud.
Considine: We’ve done a lot of work on our cloud to disaggregate the infrastructure. The first level of disaggregation is that the storage is not captive to a specific server. By disaggregating it, as you build your server or your storage for your container, or whatever else, then storage isn’t bound to a specific computer. That’s goodness. Because if that computer fails, you don’t lose the storage with it. By putting it out there, it’s available to be constructed against needs as it evolves in the infrastructure.
The next step: the idea of devices. GPUs [graphics processing units] are a prime example, and all forms of accelerators. Instead of saying I’m going to put a big box and plug 16 of these or 32 into a box, I’m going to put the GPUs out on the fabric. Then if someone comes to me and says I’m building this really crazy thing and I want 256 GPUs plugged into this CPU, I don’t actually have to have a box with 256 of them. I can actually aggregate them across the fabric and logically place them together.
Now the next step is why don’t we do that for memory? Why don’t we do that for all resources? Generate it into pools and then construct it according to the specific applications. That carries forward the real dynamicism in cloud. Part of the dynamic nature [is that] we can increase the resource density during active periods and then we can decrease it during passive or different levels of consumption. Big benefits: boost utilization, boost efficiencies, lower cost.
Photo: John Considine, GM of IBM Cloud Infrastructure, during the opening keynote at Cloud Expo 2017.