I have been following the cloud computing discussion group for a while, and even see some people with important sounding titles from important companies put up their “vision” about cloud computing, etc. with very little substance technical or business wise in their presentations. One then wonders why folks like Larry Ellison say that there is nothing new here.
What I do not see is a simple explanation of what is the change that cloud computing engenders for IT and what are some of the challenges for cloud computing to succeed in supplanting the current model of IT that is prevalent in the industry. People keep talking about how the existing business models will be busted, disrupted, etc. without necessarily having all the evidence. Just because one asserts that the current model will be broken, does not mean that it will.
I have been quoted in the past as saying that technical progress should be evolutionary, not revolutionary mostly because people who use these technologies want to evolve, not revolve. 🙂 (http://www.usenix.org/publications/library/proceedings/worlds04/tech/full_papers/karp/karp_html/index.html) I think the same is true of cloud computing.
Seriously, there is a real danger of us taking on the wrong problems in cloud computing and it will end up not achieving its potential. Some of these thoughs were spurred by my old friend Krishna Sankar’s blogs and posts on cloud computing (http://doubleclix.wordpress.com/2008/10/21/what-is-cloud-computing-and-do-i-need-to-be-scared/), and the active discussion on cloud computing group.
I wrote this piece for Krishna Sankar’s blog, but decided to make it my first stab at articulating what I think the axes are along which cloud computing claims to change the current model of IT, and what are some of the challenges that will have to be overcome, and what are the ‘limits’ of use for cloud computing.
I like to think of things along three dimensions for managing IT environments:
People (can be shared or dedicated),
Technology (can be standard or custom), and
Processes (can be manual or automated)
Now, you may say there is a continuum along all these dimensions, and you are right. But, go along with me for the moment..
Cloud computing implies: shared resources, standard technology, and automated processes. Amazon’s EC2 does not have folks dedicated to my account, it gives me a fixed set of standard technologies, and the provisioning/management is highly automated.
Most IT shops at an aggregate (there are clear exceptions, but I am trying to make a point here) are closer to the opposite end of the spectrum, i.e.,
People are likely to be dedicated
Technology is likely to be custom
Processes are likely to be manual
Many IT shops have some notion of shared services, but more often than not, there are dedicated teams per organizaitonal unit, even it the IT as a whole is centralized. The technology decisions are made wherever they are made, and often each IT shop has a variety of technologies through out the stack, and the processes are often not even standard, i.e, ITIL compliant, let alone automated.
Outsourced service providers maybe better than vanilla IT departments in some areas such as adherence to standard processes, but the people and technology dimensions are unlikely to be very different. Outsourced service providers would like to standardize the technology where possible, but, often they have to support what the customer currently has, and the cost of moving technologies to the ‘standard’ platform is often difficult to justify for both the customer and the service provider.
In theory, cloud is an enabler, for both in-house IT Shops and outsourced service providers to dramatically reduce the operating cost and move to the opposite end of the spectrum.
In theory, there is nothing ‘fundamentally new’ in cloud computing if you abstract it enough. However, the actual realization can cause dramatic difference on how IT solutions are created, delivered, managed, etc.
In theory, one analogy that I find interesting is the airline one, especially Southwest. It is not an accident that Southwest is more profitable than any other airline, in no small part because of standardizing on 737s. However, the flip side of it is that they dont do long-hauls, esp international long-hauls.
So, the point I want to make is that cloud computing may work well for some parts of the IT, and not so well for others. Parts that are likely to be standard across customers, dont have sensitive data elements that one has to demonstrate control over for compliance purposes, etc.
Here are some things to think about though..
1. If you are a company that already has a lot of investment in IT, you have many applications supporting many business processes, and you are looking to get the same ‘functionality’ but at a lower cost point. That is you have a ‘defensive’ strategy as it relates to IT, or to put it another way, the business and functional requirements dont change (or at least dont change a whole lot), but you want the technical and implementation components to transform to provide you a new cost point, while cloud computing seems to fit that bill very well, the challenge is in transforming the existing IT hair-ball that the customer likely has into the standards-based cloud platform. Of course, there are specific areas in your IT environment where this transformation is easier, e.g., messaging and collaboration, and maybe that is a good foothold for cloud computing to take hold. But, i dont think one should generalize to all of IT before we can at least get a clear picture of the problems we will need to solve.
(If you have only 737s, you can do only short hauls. If you want to do long hauls, maybe you need to have A380s)
2. If you are an outsourced service provider who has taken over someone’s ‘mess for less, and you think that you can do the same, you will have the same issues. The issue is transformation. And more importantly, who pays for it. So, unless cloud computing providers provide ‘on-ramps’ for migrating onto the cloud platform, they may severely limit the impact of cloud computing on existing IT solutions.
3. If you are an IT shop that is willing to invest in new capabilities, either because there are processes that you can ‘automate’, or because you can open up new revenue possibilities through technology, then the cloud platform is a platform definitely worth considering from the beginning. The problem in this case is only integrating this specific ‘cloud app’ with other things in your enterprise, and that is a relatively easier problem to solve. So, cloud computing folks should focus on creating the next generation of these apps. Anticipate the next gen of IT spending and enable the alternative economics on that.
4. If you are an outsourced service provider, one way you can ‘innovate’ is by enabling these new apps on the new platform, else all you will be doing is ‘your mess for less’, squeezing squeezed lemons, trying to make lemonade, etc.. 
Most clients expect outsourced service providers to expose them to the new platforms, and shield the client from technical disruptions. And it is indeed the responsibility of the service provider to do that.
Kannan's Blog 