David Chappell


Get the Feed! Subscribe

Why It's Hard for Enterprises to Save Money with Cloud Platforms  
# Thursday, January 22, 2009
There's a widespread belief that once a company moves its in-house applications onto a cloud platform, its costs will go down. Intuitively, the idea makes sense. After all, big cloud providers like Microsoft and Google should enjoy incredible economies of scale. (I don't care who you are--their data centers are bigger than your data centers.) And assuming these economies are reflected in their prices, running your apps in their cloud ought to be cheaper, right?

Well, sure--someday. If you look at what the most visible cloud providers offer today, however, this assumption turns out to be largely incorrect. The primary falsehood is the belief that an organization can move its on-premises applications to the cloud unchanged. While this isn't impossible--more on that in a minute--it can certainly be problematic.

Consider Google's AppEngine, for example. It supports only one kind of application: Python Web apps. How many existing enterprise apps fit this description? And even if you do have an on-premises Python app that you'd like to move to the cloud, AppEngine's datastore is a Google-proprietary hierarchical system--it's not a relational database with SQL. Moving an existing application to this world, however cheap it might be to run once it's there, isn't an option for most.

Microsoft's Windows Azure offers similar challenges (today, at least). While it does provide a reasonably standard .NET environment, it also offers only a proprietary hierarchical datastore--no relational tables and no SQL. Once again, moving existing apps to this world is more than a little challenging.

In fact, neither AppEngine nor Windows Azure today is intended to run existing on-premises apps unchanged. Instead, both focus on letting us build very scalable applications for the Web. This is certainly a worthy goal--it's exactly what you need for building some new software. But it's not much help to an organization that just wants to run its current apps more cheaply.

Of the most visible cloud platforms today, Amazon's EC2 is the only one that offers the ability to run on-premises applications in the cloud. With EC2, you really do have the potential to move your apps to Amazon's cloud unchanged. Even then, though, will you save money? It depends, as others have pointed out. And other complexities also emerge, such as the challenge of managing applications running on a cloud platform. Even with EC2, moving your current apps into the cloud isn't a no-brainer.

A related point: It's often claimed that cloud platforms can provide a cheap on-demand solution for handling extra computing load, much as electric utilities rely on gas generators to handle spikes in demand. But this is only possible if I can run the same app unchanged in the cloud and on-premises (or, I suppose, if I've got a version of the app and its data for each environment, which isn't a simple thing to create). Once again, I can do this with EC2 but not (today, at least) with AppEngine or Windows Azure.

The equation seems simple: Cloud platform economies of scale equal a cheaper way for me to run my current apps. Yet when you look at what's out there today from the big vendors, much of what's on offer just doesn't allow this. Wanting something to be true doesn’t make it so.

6 comments :: Post a Comment



I agree the custom-app-port-to-cloud isn't really viable today. However, moving commodity software products out of a local data center and into the cloud has the potential, IMO, to save a lot of money: hosted Sharepoint, Exchange, etc. I think it will be 2011 before we start to realize custom app benefits in the cloud.

Very nice post.. I think it will take 3-4 years before cloud platforms are mature enough for enterprises to give them serious consideration..But I think the ladder is leaning against the right wall.. Cloud platforms,from and only from big vendors like Microsoft/Amazon/Google, is hear to stay

Focusing on the cloud and the "Enterprise" misses the true benefit the technology will have on society.
Cloud platforms are a competitive advantage for new innovative startups interested in building Software as a Service solutions. There are exceptions, but this seems to be the current sweet spot for the cloud.
It will be interesting to see the types us services that will come out of cloud computing. I smell innovation! Giving the small development shops so much computing power with little upfront cost will be a good thing for all.

I pretty much agree with all of these comments. The last one is especially interesting: Web start-ups clearly are a major user of cloud platforms, and easy access to cheap data centers is a great thing for them (and through the innovations those start-ups bring, for us as well).

Still, if start-ups were to comprise the primary cloud platform user base for the indefinite future, I wonder how long these new platforms can last--start-ups aren't a big enough market to sustain much. The big IT money remains in enterprises, and so my hypothesis is that cloud platforms must attract enterprise applications to be long-term viable.

To the point that Enterprises have the only funds necessary to sustain cloud computing as a business model...Not all start ups remain small. Take a look at YouTube, Del.icio.us, Flickr, etc. There are also startup companies that will service the enterprise with innovative cloud based technology. The Enterprise may not innovate in the cloud, but Enterprise money will be paying for the cloud by consuming startup technology. At least until cloud is a trusted and accepted by many enterprises as a way to deploy technology, which IMHO will take awhile. Which isn't necessarily a bad thing, it just represents an opportunity for small companies.

Thanks for the post. You focus your attention to the development part, but the production side of it is also an issue. Large enterprises will not be able to rely on a sole cloud provider, so they will still have their traditional IT (outsourced or not). As I described in http://buzina.wordpress.com/2009/02/05/the-cloud-cost-saver-or-brain-saver/ there are other factors often neglected which drive the costs of such an endevor.

Post a Comment

<< Home