Tuesday, April 13, 2010

Forrester's Staten: Realities of private and hybrid clouds aren't what you're expecting

James Staten does not pull punches. And for an IT industry analyst, that’s a good thing.
I first met James a few years back when he joined Forrester during my time at Cassatt. I heard him do a couple presentations at that year’s Forrester IT Forum, had some briefing sessions with him, and realized that with James, friendly conversations quickly turn into very specific advice and commentary. Even better, it was advice and commentary that was very much on target.

For those of you who don’t know James, he is a principal analyst in Forrester Research’s Infrastructure & Operations practice, helping IT ops folks make sense of topics like virtualization, cloud computing, IT consolidation best practices, and other data center- and server-focused issues. From my perspective, James has been a great addition to Forrester, especially in his role as one of the earliest voices helping describe the impact and meaning of cloud computing.

So, I thought I’d turn him loose on a couple current cloud computing topics and see if we couldn’t find a few things to argue over. In a good way, of course. Here’s that interview:
Jay Fry, Data Center Dialog: There’s lots of debate about how cloud computing takes hold in an organization. Some see people starting with public clouds. There is also lots of conversation about private clouds. In your discussions with customers, what path do you see end user organizations taking now – and is it even possible to make any generalizations?

James Staten, Forrester: Most of the time enterprises start by experimenting with public clouds. They are easy to consume and deliver fast results. And typically the folks in the enterprises who do so are application developers and they are simply trying to get their work done as fast and easy as possible and see IT ops as slow, expensive, and a headache to deal with.

In response to this, IT ops likes the idea of a private cloud – their word, not mine. This usually translates into an internal cloud and the desire is to transform parts (if not all) of their virtual server environment into a cloud. Usually this transformation happens only in name, not in operation -- and that’s where a disconnect arises. IT ops pros tend to rebut public cloud by saying, “Hey, I can provision a VM in minutes, too.” But that’s not the full value of cloud computing. Fast deployment is just the beginning.

DCD: How do you see hybrid cloud environments coming into the picture? How soon will that be a reality?

James Staten: It’s a reality for some firms today, but not in the way that many people think. A hybrid between your internal data center and a public cloud isn’t very realistic due to latency, cost, and security. A hybrid between your internal cloud and a public cloud isn’t realistic either. (See the above answers for why.) What is realistic is a hybrid between dedicated hosting and cloud hosting within the same hosting provider. USA.gov is doing just such a thing in Terremark’s Alexandria, VA data center (Forrester customers can read a case study on this here). This kind of a hybrid allows the two separate environments to share the same backbone network and security parameters. And it allows the business service being supported to match the right type of resources to the right parts of the service.
For example, you may not want or need elastic scalability for your database tier, so it’s more stable and economical to host it on a dedicated resource with a predictable 12-month contract. But you have a lot of resource consumption volatility in the app and web tiers and so they are best served being hosted in a cloud environment.

DCD: I’ve written here before about rogue deployments using cloud computing from business users without the explicit buy-in of the IT department. Do you see that as likely or commonplace? How should IT deal with these?
James Staten: They are extremely common place as evidenced by our role-based research, which shows that when you ask application developers if their company is using cloud, you get back that 24% say yes; you ask the same question to IT ops pros and get back that 5% say yes. Clearly app dev is using the cloud and cimcumventing IT ops in doing so. It’s no surprise. They see IT ops as slow and rigid.

How should IT ops respond? First, you can’t be draconian and say, “Don’t use cloud.” Those who are parents know how well that works. Instead, IT ops needs to add value to this activity so they are invited into this process. They need to embrace the use of public clouds by asking how they can help.

DCD: What are some of the big changes that you see underway with regard to cloud computing this year so far? Has anything really surprised you?
James Staten: This year will be all about understanding where to start for most companies, how to move from testing the waters for the percentage who have already gotten this far, and how to optimize your cloud deployment for those who have already moved into production on public clouds.

For IT ops, this year is about roadmapping your transformation for running a virtual server environment to running an internal cloud. They won’t get there in one year but they can build the plan and start moving now.

Nothing has really surprised me in the past 12 months but I look forward to seeing the innovative ways companies will devise to take advantage of clouds in the future. We’ve seen some very promising starts.

DCD: Acquisitions are shaping up to be a major storyline this year from my point of view. And I’m not just saying that because of CA’s recent moves with Cassatt, Oblicore, 3Tera, and Nimsoft; there have been many. You were recently commenting on Twitter about confusion coming from Oracle about the management products related to their Sun acquisition. What role do you think acquisitions are (and should) play in shaping what customers have to choose from in the cloud computing space?

James Staten: For the most part, acquisitions in a new emerging space are about speed to market. The leading software companies could adapt their existing enterprise products to incorporate and integrate cloud computing, but that’s hard, as there are already 50-100 other priorities on the roadmap that keep the installed base happy. Acquisitions also inject new blood (new thinking, new technologies, perhaps even new culture) into an existing player and that is often sorely needed to drive a sense of urgency around addressing a new market opportunity because the immediate revenue opportunity in the new market is much smaller. This is at the heart of Clayton Christensen’s Innovator’s Dilemma. Cloud computing looks very much like a disruptive innovation and according to Christensen’s theory, the disruption can be a king-maker for those that lead the disruption (and make a pauper out of those being disrupted). So that fuels the belief that acquisitions are necessary.

DCD: You wrote one of the earliest analyst reports on cloud computing in March 2008 (“Is Cloud Computing Ready for the Enterprise?”). You’ve also been on top of “how to” topics for end users with research to help clarify some of the fuzziness around cloud computing (“Which Cloud Computing Platform is Right for You?” [April 2009], “Best Practices: Infrastructure-as-a-Service” [Sept. 2009], etc.). Given how the cloud computing category – and the industry debate – has evolved, would you approach some of your earlier research differently now? Any conclusions you would change?

James Staten: No, I feel proud of the research we have done as it was backed by real-world interviews with those who are leading this evolution, rather than theories and leaps of faith about what might occur. We were also very clear in our objective of clarifying what was truly new and different about cloud computing to help guide customers through what we knew would be a hype-filled world. It’s unfortunate the industry has latched on so tightly to the term “private cloud” because in our opinion it is a very nebulous and thus meaningless concept. Heck, anything can be made private. But then again, cloud computing isn’t most precise term, either.
DCD: If many years from now we’re thinking back on how cloud computing got started versus how it ended up, there will be an interesting storyline about the degree of difference between the two. How much of the hype around cloud computing should be ignored and how fundamental do you think it will end up being? What’s going to be the thing that really makes cloud more mainstream in your opinion?
James Staten: The one part about the hype that I think should be ignored is the belief that everything will be cloud in the future. While this is a nice, disruptive statement that draws a ton of discussion and "what if" thinking, it simply isn’t realistic. If it were, there would be no mainframes today. Everything old is still running in enterprise data centers; we simply contain these technologies as we move to adopt new ones that arise.

And cloud computing is definitely not an answer to every question. Some applications have no business in the cloud and frankly will be less efficient if put there.

It’s through better understanding of what the cloud is good for and what it isn’t that IT will move forward. At the heart of cloud computing is the idea of shared empowerment – that by agreeing to a standardized way of doing something, we can share that implementation and thus garner greater efficiencies from it. That concept will manifest itself in many ways over the coming years. SaaS is a classic example that is delivering new and greater functionality to customers faster. IaaS is a great example because when multiple customers share the same pool of resources the overall utilization of the pool can be higher and thus the cost of the infrastructure can be spread more effectively across a bigger customer set, lowering the cost of deployment for all. Take this further and you can imagine many more scenarios:
· Why buy any software when it is more efficient to rent it only when you need it? We haven’t seen that model in SaaS yet.
· Why write code when you can more easily construct a business service by stringing service elements together? This is the core of SOA and builds upon the Java construct of reusable libraries. How far can we take that?
· Why store anything yourself when storing things on the Internet allows that data to be anywhere you are, whenever you need it to be – and to be self-correcting? I’ve been using Plaxo for over 5 years to keep track of all my business contacts so I don’t really care if I lose my mobile phone, laptop, or other device that stores this information. Now that Plaxo links with Facebook and LinkedIn, all the people in my address book can update their own records and I get this info as soon as it is synced. That’s distributed data management in the cloud. Can this same model be applied to other data and storage problems?
DCD: So, what area do you think is the next battleground?
James Staten: There are many areas that will be effected by cloud computing. One market that greatly interests me right now is HPC/grid computing. Loosely-coupled applications are a great fit for the cloud today and flip the economics of grid on their head. There are some incredible examples of companies using this combination to transform how they do business in healthcare, financial services, government and many other fields. Business intelligence is fertile ground for change due to this and the influx of MapReduce. I’m really looking forward to seeing what changes in this arena.

DCD: What do you believe will be some of the more interesting starting points for customers’ cloud-based activity?

James Staten: You gotta start with test and development because this is the one area where any company can get immediate benefit. Every enterprise has a queue of test projects waiting to get lab resources. Use the cloud as an escape valve for these projects. Then take a look at your web applications. If they have traffic volatility they are natural fit for cloud.

DCD: Do you see any dead-ends that some customers are heading down that others should be careful to avoid?

James Staten: It’s a total dead end to think that you can simply buy a “cloud in a box” and suddenly you have an internal cloud. Internal cloud isn’t a “what,” it’s a “how.” How you operate the environment is far more the determiner of [the benefit of] cloud computing than the infrastructure and software technologies it is based on. This is true for service providers, too. Just because an ISP has expertise in managing physical or virtual servers doesn’t mean they can effectively run a cloud. Sure, some of the cloud building block technologies can help you get there, but this is totally an operational efficiency play.

DCD: Forrester recently changed its blogging policy for its analysts to require that any content in research-related areas be posted only to official Forrester blogs. Did this move make it harder or easier to blog, and do you think it’s going to help customers in the long run?

James Staten: It made it significantly easier for me as I strongly believe in separation between work and personal life, and equally believe in freedom of expression – and thankfully, so does Forrester. While I can’t speak for every analyst (nor can I speak for Forrester on this topic), I can say personally that this change in policy was a good one. Prior to this we had team blogs, rather than blogs for each analyst, and if someone wanted their own stage, they had to go outside to do it. Now our blogging platform gives every analyst their own outlet while preserving the aggregation of blog content by client role. We also moved to a blogging system that makes it much, much easier for me to author and publish blog entries myself. Anything that makes me more productive and helps clients consume our value is a good thing.

Thanks to James for spending the time for this extended interview. Of course, given that James is a pretty intense endurance runner during his off hours (he’s shooting for 8 marathons and 6 half-marathons this year – and 50 marathons by his 50th birthday), a marathon interview seemed somewhat appropriate.

If you have thoughts or feel an urge to disagree with James about any of topics we touched upon here, feel free to add your comments.

No comments: