Ann All spoke with Ross Tisnovsky, vice president of research for Everest Group, the author of a recently published study titled “Hype and Reality of Cloud Computing – Mind the Gap.” Everest Group uses proprietary market knowledge, frameworks, skill sets and insights developed over the years to offer a flexible portfolio of consulting and research services.
All: Everest finds that cloud computing benefits are less compelling for companies that have already virtualized their infrastructures. For these kinds of companies, do some variants of the cloud make more sense than others?
Tisnovsky: If you look at cloud computing, it offers a lot of benefits by sharing the same platform, the same servers or mainframe across multiple buyers. That is a huge source of benefit. But there’s additional overhead in cloud computing. The supplier needs additional software, they need to build an engine, there are a lot of challenges around maintaining service-level agreements. Those overheads can kill the benefit of cloud computing in the virtualized environment. When you virtualize your internal environment as a buyer, you do essentially what the cloud computing supplier is trying to achieve. You put multiple applications on the same platform. So some of the benefit that cloud computing offers will already have been taken, and additional overhead comes in.
Having said that, there are extreme cases of buyer demand which, even under these conditions, will derive benefits from cloud computing. An example is testing. In testing, it’s not that you want to leverage one platform across multiple applications, it’s that you suddenly need the platform, and the next day you don’t. So of course, having the opportunity to have the platform one day and not have it the next day will make cloud computing a valuable solution. This prompts us to ask a question: Are we going to have a wide adoption of cloud computing, or will we have niche-based adoption of cloud computing?
All: Right. I believe you mention three adoption scenarios in the report: niche adoption, industry consensus and hype and decline. Which do you see as being most likely? Why?
Tisnovsky: I am biased toward the niche scenario. I think niche-based adoption makes a lot of sense because we can identify a large number of niches, all of them extremely different. First I’d say niche by size. Large enterprises aren’t likely to go 100 percent in the cloud, but SMBs might. Another niche is extreme demand, like testing. Another niche is industry-specific cloud applications. There are multiple suppliers attempting to build those, and health care seems to be the favorite. There aren’t too many businesses that have requirements similar to those of, say, a pharmaceutical company doing clinical trials. They have very specific needs, and if you build a solution that satisfies those needs and also complies with all of the health care regulations, that would be a very specific cloud.
All: In your report, you also discuss operations-as-a-service, which you say is still immature but presents the most opportunity in the infrastructure layer of the IT stack. Can you elaborate on this concept and how you see it developing?
Tisnovsky: If you look at the four layers that exist today – there’s business-process-as-a-service on top, software-as-a-service as the next layer, platform-as-a-service and then infrastructure-as-a-service on the very bottom. So let me give you a cynical assessment of those. Business-process-a-service, we’ve probably been doing for 20 years. Any business process outsourcing, or BPO, is by definition conducted from the cloud. Look at ADP payroll services. ADP has been around for years. Their customers didn’t know they were using the cloud until guys like me came forward and said, “Yes you are using the cloud.”
Software-as-a-service is an interesting and viable solution. Salesforce.com has been around for 11 years, I think. It’s an ongoing phenomenon. I don’t get the big deal, if we are now calling it the cloud.
Platform-as-a-service is an interesting and smart concept. Look at Microsoft’s Azure, for example. What it does is give you a platform to develop your applications, then test them and host them right there. It’s a very convenient way of building new applications. The key words are “new applications.” You cannot do it with your trading platform if you’re an investment bank that’s been around 20 or 30 years. There’s no way you can move all of your applications to a new platform. If you could, you would have done it 15 years ago. For most enterprises, new applications can be from 3 percent to 15 percent of the IT expenditure. So it’s a good niche solution, maybe 10 percent of your spend.
That gets us to the infrastructure-as-a-service. Quite honestly, that’s where all the problems are. Because if you want to carve an application out of your infrastructure, an application with the operating system and hardware underneath it, and you want to put it in the cloud, you’re going to have a significant problem with integration, with privacy, with security, with regulations and with data management.
So think about it. What if you don’t touch that server? What you deliver from the cloud is just the administration of the server, the management of it. There are no data-management concerns because data still resides with the buyer. What is delivered from the cloud is the operations layer, which are the people.
All: So it’s a resources play? You save on staffing and access best practices that you may not have had in place in your IT organization?
Tisnovsky: Correct. But don’t confuse it with staff augmentation. With staff augmentation, you provide bodies working on certain servers. What you provide with operations-as-a-service is an actual service which is charged directly per server, per terabyte of storage, per any measure of output. It’s flexible and can go up and down, depending on how much you consume.
So it has all the characteristics of the cloud: pricing based per any measure of computing output; built so it’s easy on, easy off; it can scale up or down. If your provider can give you flexibility, it behaves like other cloud solutions. Hence we argue there is a cloud solution called operations-as-a-service that is largely missed by suppliers and buyers today.
All: Are any companies offering approaches you think come close to this concept?
Tisnovsky: We see some of it. Cognizant is one company that has modular solutions. Perot, before the acquisition by Dell, had modular solutions. The reason modularity is important is because that’s what makes it easy-on, easy-off. But it’s not mainstream today.
All: What will it take to make the concept more mainstream?
Tisnovsky: It’s contingent on the sophistication of the buyers. Despite the easy-on, easy-off nature of it, it will require management on the buyer’s part, and better governance. They’ll now have multiple suppliers managing the server, with maybe one providing a modular monitoring service, another modular problem resolution, a third one doing modular backup.
All: The report references five strategies employed by cloud providers: adviser, reseller, orchestrator, enabler and engine. Which of these are currently most popular? Do you think this will change in the future?
Tisnovsky: Everybody is an adviser. Some companies have more sophisticated systems, with intellectual property behind it. TCS has systems that help you test your applications before putting them in the cloud and identify cloud-ready applications in your portfolio. I do believe there is IP around it.
Engine is the least populated one. Amazon’s Elastic Cloud kind of has a retail spin on it. IBM and HP built clouds, which makes sense because they also sell hardware. OEMs are definitely interested in it. Former hosting companies are interested. There are a limited number of companies positioning themselves as engines.
I think the most valuable ones are enabler and orchestrator. Security needs to be orchestrated. If you’re using multiple clouds, you need a comprehensive security solution. There are companies positioning themselves as orchestrators of security. They are companies positioning themselves as orchestrators of data. CSC is a good example. They have a data solution that I believe comes from their government legacy. Data is a nightmare in the cloud. It’s not always easy to locate your data in the cloud.
Capgemini is probably a company that stands out as an enabler. They have partnerships with Google and with Amazon, and they actively resell and enable those solutions. The thing that not many buyers understand, if you go and get resources from Amazon, they do need to be managed.
All: You mention two drivers of adoption: technology-led (cloud interoperability and security are given as examples) and services-led (migration services and management/monitoring). Are suppliers addressing these areas simultaneously, or is one getting more attention than the other?
Tisnovsky: There are companies in the services space that move into cloud computing, and they definitely address both. For technology companies such as software companies or data center hosting companies, it’s much more difficult to move into the service area.
All: Most of the discussions I hear about cloud computing focus more on technology than associated services. Why do suppliers seem to be focusing more on technology?
Tisnovsky: Because that’s where the value is. If you look at cloud computing and really simplify it, the No. 1 driver of value is utilization of hardware. You utilize your hardware better by sharing it with multiple buyers.