If you are a little fuzzy on what cloud computing today actually means, this interview with IBM cloud expert Bruce Otte may prove enlightening.
IBM issued a news release several months ago that reported the results of a survey measuring system administrator preferences regarding cloud computing services. As it turned out, IBM's cloud services were preferred over those of other providers, the survey found. We began to reflect on the terms "private cloud" versus "public cloud" mentioned in the release and realized the definitions weren't clear to us. So we invited Bruce Otte, IBM senior marketing manager, Enterprise Initiatives, Cloud Computing, to elaborate on the different types of clouds. The results are quite revealing, since cloud technology is still evolving. Following is the first in a two-part edited interview, conducted last August but published here for the first time, in which Otte explains the current state of cloud computing and the jargon that has grown to describe it.
Chris Smith: Bruce, could we get your official title and what your responsibilities are?
Bruce Otte: I'm Bruce Otte, and I'm part of the IBM Initiatives Team. I do cloud computing marketing and market strategy and have been working on cloud for two years. Prior to that, I was in dynamic infrastructure, service management, and SOA—all three things that are keenly related to effective and efficient cloud delivery.
Smith: I have you down as senior marketing manager, IBM Cloud Computing initiative? Is that what they're calling it these days? Is there a more formal term for the department?
Otte: It's Enterprise Initiatives, Cloud Computing. We…recognize that in the future there will be other enterprise initiatives as well; this is just the first big one.
If you want to talk about a maturing model, it started out with very centralized computing in the '60s and '70s and turned into just very distributed computing in the '70s and '80s, but at the end of the day, people really didn't want a computer; what they wanted was access to what the computer could do. And that's led us into the first decade of the 21st Century into this thing called cloud computing, where everybody still has their distributed devices, but they just simply get access to the things they need through the cloud.
Smith: The thing that I was a little confused by, and what led to this call, is this term "private cloud" as opposed to "public cloud." There was a release that IBM put out that said the IBM cloud was the preferred private cloud of system administrators, and it sort of took me back a little bit. I was thinking that if people were contracting with IBM for cloud services, then isn't that a public cloud? I always thought of a private cloud as being within the corporation and set up offering services to the employees—not something that would be sold to companies or individuals on the outside. That was one point that I found a little confusing. So, is the IBM cloud a public cloud or a private cloud?
Otte: That's a great question. It really goes to how you define public and private…. But, if we have the fairest of definitions, a private cloud is contained inside your four walls, your data center, exclusively for an audience that you designate—typically, employees. It may expand to business partners or those you're doing business with, but it's all something contained inside your four walls and only for your internal users. A private cloud typically doesn't do billing…or pay per use, but may do some usage tracking and charge that. It's still a cloud delivery model enabling the end users to be in charge, and the resources are typically shared, but it's all inside your firewall. If you visualize a picture [depicting cloud computing, imagine the private cloud] on the left.
On the far right…side of this picture is this thing called "public cloud." The barest definition of public cloud is the exact opposite of [private cloud]. It's a third-party provider service, it's open and completely shared—all the resources are totally shared—and it's open to anybody. It's pay per use. Payment may be monetary or may be [paid] in other ways. For example, with Google's public cloud, your payment is your information in that every bit of information that goes across their network is data mined, and they pull all that data-mining together, and it's that information that they sell to advertising and other sources.
Whether you're paying $3 per seat for a collective number of seats on their Gmail or whether they are getting their Gmail free, in both cases every bit of information you have is data-mined. They supplement that $3 per seat by mining that information as well. [Payment] may or may not be monetary, but [the model] is typically a very shared infrastructure. It's very much multi-tenant in that everybody is accessing the same application, same resources simultaneously.
If you picture that [public cloud] on the far right…that pure public model, [and] if you come in one square [from] the right of that, there's this thing called "community services"—some call it shared services, others call it member services. The idea is that it's a third-party provided service, but you have to be a member or a participant of the community to be able to access that service. That's where IBM's offering sits. So, we don't call [it] a pure public offering because you can't just come in and swipe the credit card and be John Doe and sign up and get resources. We need to provide a better level of security and protection for our enterprise members, and so our services are what we call "member services" or "shared services." It's one box in from the right.
Now, if you look at the center of the picture, you could have in that box what we call a "hosted private cloud." So if you have the private cloud on the far left, [in the center] you can have a "hosted private cloud." That's where it's still a private cloud, but it's not inside your data center; it's inside a hosted center. It could be one where you manage it or that third party manages it, but all the resources are dedicated to you. You pay for it in more of a traditional hosting manner, but it's still for delivering those cloud-focused services. One box in from that, between the far left public and the hosted would be what we coined [the] "managed private cloud" where, yes, it's inside your data center, but the third party is the one responsible for completely managing it and putting it together. Those are what we call the "deployment options cloud."
Smith: That's interesting; I don't know how many people fully understand the different gradations.
Otte: I don't think they do. It's new to the market. It's something we started introducing last year. IDC, Gartner, and Forester—the three analyst firms—are also talking about it in their charts. In effect Gartner's…chart shows [a model] very similar to [ours]…with the private, managed, hosted, shared, and public [clouds]—those five gradations. But it is new because cloud is new; this whole thought process is new as well. The nice thing is that we're getting away from the thought that you buy cloud and into the thought that cloud is really a delivery model or a delivery philosophy that says that I want to put the end user in charge of their experience and what they're going to do and how they're doing to do it.
Smith: One of the big questions that keeps coming up, and one of the concerns that CIOs have, is this one about security. I gather that there are different security options available to each of these five levels or gradations?
Otte: There definitely are. Within cloud security, the three biggest concerns that firms tend to have are: Concern number one is "I'm not really exactly sure where the information is located, so how can I trust where it's located?" That concern is going to be addressed in a couple of ways, and one of them is going to be, well, "Who's your provider?" Is your provider a trusted partner? Is your provider an unknown partner? And the second is, what level of visibility is the provider going to give to you as to what is going on in the environment so [that] you know what's happening with your data and your information [while] your people are out there working on it? So that's one of their big concerns, that whole control factor if you will.
A second big concern on security obviously is an unfamiliarity with virtualized environments where you have hypervisors and several servers running on a physical device simultaneously so this data is running simultaneously. And the concern about hypervisors is—we really tried to address from a pure technology standpoint—the hypervisors today across all the platforms are at least—there are security levels, and I forget the terminology for it, but I believe that there are security levels one through five. All the hypervisors are rated at least a level 3, and even the mainframe System z hypervisor from IBM is rated at a level 4. A level 5 [means] the server is totally secure; it's totally inside your environment and has no Internet connections whatsoever. It's extreme, from a level 5 standpoint. But a level 4 is a very high security rating, and the hypervisors are at least at a level 3. So it's typically the hypervisors tend to not be the issue; the issue tends to be what data are you running, when are you running it, and how can you provide audit ability over what's occurring—when, where, and how. For some types of data, particularly customer data that's going to contain private information, or employee data, or in the case of anyone associated with the healthcare industry, patient data. Those are three sets of information that…at any point in time you have to be able to indicate where the data is, who has access to it, how that data is secured, how it's encrypted, and all those sorts of things, and when you're dealing with a cloud delivery model, where you're virtualizing a server. You can always tell it's on this virtual server, but you may not know which physical device it's running on. It could be spanning devices that give you your virtual server. That's what presents some of the issues and concerns around [the] security side of it. It still is today when you start talking about going into the public cloud….
Smith: In an audit and in a virtualized environment, are you expected to know which physical server the data resides on, or is a virtual server good enough?
Otte: It depends on the type of data. For patient data, you're supposed to know exactly which device and what location. They have not updated [auditing procedures] to account for this thing called "cloud." … [Regardless of your provider], as the owner of the content, [you] are ultimately the one responsible for the audit. That's one of the reasons in the banking industry we're seeing a number of banks migrating over to what we call a desktop cloud. Because in the past, where they had these distributed individual desktop servers of laptops—notebooks for employees—there was customer data residing there.… Audits are getting stiff enough now that they have to know what customer data is residing where and who has access to it. There was a lot of concern about [saying] "we don't really know." But if we put it into a virtual desktop, then we know it's sitting in this device, in this data center, and we know that's where the data resides, and the only one who has access to that are these XYZ employees.
Smith: They're calling that a cloud too?
Otte: A desktop cloud is basically a virtualized desktop environment, and you can have private desktop clouds. It is called a cloud because once you build out the image, it's the end user who is in charge of what happens when, where, how—and all you do is maintain that form in the background. That's what makes it a cloud delivery model.
Next: IBM's shared services model cloud and who is adopting it.
LATEST COMMENTS
MC Press Online