ESG's Terri McClure and Dan Conde discuss Dan's thoughts and predictions on Cloud for 2018.
Read the related ESG Blog: ESG Blog: 2018 Predictions for Cloud Platforms and Services (Video)
Dan: Hi. I'm Dan Conde, analyst for cloud platforms and networking at ESG, and...
Terri: And I'm Terri McClure, a senior analyst for cloud infrastructure at ESG.
Dan: So, here we are, looking at what's going to be important in 2018 and some predictions of what may be critical things that enterprise customers have to think about. So Terri, you have some questions for me?
Terri: I sure do. There's a lot going on in your space this year. The things I hear all the time are all around hybrid cloud's gonna be the future, multi-cloud's gonna be the future. People aren't happy just using a cloud, whatever a cloud may be. Can you elaborate on...do you think this is a reality? Are people really adopting many clouds and hybrid clouds?
Dan: That's a good question. And it's true that people think about the cloud as one thing, but in reality, our research shows that people are using multiple clouds now and even intend to use multiple clouds in the future. And that can also mean off-premises clouds as well as on-premises equipment, too, so the combinations are quite complex.
Terri: So private plus what we think of as public cloud?
Dan: That's right.
Terri: Is this...but when I hear about hybrid cloud or, I'm sorry, multi-cloud strategies, are we looking at people using infrastructure as a service across multiple clouds? SaaS or is this PaaS? Is it this combination of using this cloud for SaaS, this cloud for IaaS, this cloud for PaaS? How is it coming together?
Dan: I think it's all of the above.
Terri: All of the above.
Dan: All of the above. So there are some enterprises who will want to use different or multiple clouds for different reasons. Some of them may say that, you know, this cloud is really good at machine learning and AI, so I want to use that, and another one because they have compatibility, let's say, with another piece of software. Like vSphere or VMware runs on one cloud, so they may want to use that combination.
But there are other people who want to use multiple clouds for a completely different purpose which is to sort of hedge their bets. So in that case, they wanna say, well, I wanna use all the different clouds for my workload so that I can mix and match. And that's difficult because if you're using it in a native PaaS way instead of just infrastructure, then each of them have different APIs and database services and so on. So now, the question is, do I have different versions of the applications written for each cloud? Or another option that they could take is to get third-party software that is a layer, like a PaaS layer or a container orchestration layer, that runs on different cloud platforms, public cloud and private cloud, and see if you could create a portable layer. Now, that means that you're now reliant on that particular vendor's software, so you may not be completely independent, but a lot of them are open-source, which means that you're really not locked in. They're available through different vendors, potentially, as well as you could just get the open-source version. So there are a lot of options available for the customers.
Terri: So it's really about the skills and the training and classes associated with learning all the different ways to manage clouds, and having incompatible applications across multiple clouds is a challenge?
Dan: Yeah, so that's a big challenge, right? Because in the past, if you're running Linux, for example, you would choose one Linux distro on your premises because I don't wanna train people on three different versions of Linux plus Windows or UNIX or whatever. But now, if you're running across three different clouds, you would need to learn three different security models, three different networking abstractions, and so on. So people have to realize that there are some costs associated with using multiple platforms.
Terri: You know, we've done research in a couple of different projects, now, where we've asked people, we've said, "How strongly do you agree with the following statement?" And it's this: "Increasing cloud adoption has made my IT environment more complex." And there's an awful lot of people that strongly agree with that statement. Everybody says, "Cloud, simplify your life. Swipe a credit card and get up and running. Cloud's gonna make it simple." But what we're really seeing is all this cloud adoption is making IT more complex.
Dan: Right, yeah. So I think people this year and maybe in the future will look for simplification in some ways, and vendors and software can help simplification. So underneath the covers, they could be doing a lot of heavy lifting to make things simple for the end-users, and I look forward for that.
Terri: Okay. What about on the networking side of the world? This is mostly about cloud platforms. Any big things coming in on networking?
Dan: Yeah, so the combination of networking and cloud is a lot of what's happening on the edge, and I think the edge is gonna be very sharp this year where people are realizing that they need a better way to segregate their workloads based on whether it's gonna run really well in a public or centralized cloud for, you know, data consistency and so on, and then they also realize that there is this speed of light issue. It takes time for data packets to go from the sensor, IoT sensors, all the way to the public cloud, and you don't wanna do that. Maybe there's too much data to send across the wire. So you wanna have some processing done locally on the edge and it allows you to get faster response time. You could do some sifting and filtering locally and applying analytics locally that allows you to respond immediately to the IoT sensors or anything else, for that matter, too.
Terri: Because if you do all your cloud computing at the edge, you're creating a tremendous amount of edge infrastructure, like, pet topic, right?
Terri: Then that data, whatever datasets you need, has to be backed up. You have to have some level of DR. So if you're in manufacturing, for example, your CAD files at the edge are probably too big to transmit to the core, but how do you protect your IP there and ensure it's retained in the event of an outage? There's a lot of challenges with the edge.
Dan: That's right. Its complexity comes up again, and we have to fix that.
Terri: Good. Thanks, Dan.
Dan: Thanks, Terri.