Tuesday, November 4, 2008

Who Cares About Open Source In The Cloud

Yesterday I wrote about the issue of vendor lock-in regarding cloud-based services and how I think developers should think about it. In that discussion, I touched on the open source strategy of cloud computing company 10Gen. After thinking about it I begin to believe that such a strategy may be a serious liability for cloud-based services.

Then, this morning I read an article By Nick Carr where he discussed the significance of open source to buyers. His thesis is that what is most important are the meat and potatoes issues around reliability security, etc. Specifically, Carr says:

We can (and will) have debates about the relative openness of Azure and AWS and Force.com and all the other "cloud platforms" that are available or will be available. And those will be important debates. But in this early stage of the cloud's development, openness means little to the buyer (or user). The buyers, particularly those in big companies, are nervous about the cloud even as they are becoming increasingly eager to reap the benefits the cloud can provide. What they care about right now is security, reliability, features, compatibility with their existing systems and applications, ease of adoption, stability of the vendor, and other practical concerns. In the long run, they may come to regret their lack of stress on openness, but in the here-and-now it's just not a major consideration. They want stuff that works and won't blow up in their faces.

This is very much in line with my thinking from yesterday. Azure is a big deal. No one is going to care about the fact that it is not open source. Basic hosting is going to become a commodity business very quickly, with Microsoft, Amazon, and Google (MAG) competing in the game of creating highly scalable services that use traditional development methodologies. Microsoft is now ahead in that game from a technology perspective. Amazon is ahead in customers, and Google, for now, is left in the dust but can obviously catch up. But I don’t see any of these guys making any of their cloud technology open source, and I don’t think it matters.

I liken open source in this space to DRM in the music business. Its one of those things that a small number of people complain about but will later be proven totally irrelevant to the rank and file buyer. We now have statistics to prove that DRM was irrelevant in terms of sales, and we are beginning to see the outlines of the irrelevancy of openness in the cloud.

The real issue here is that small companies are not going to be able to compete selling basic “get your applications into the cloud” type services. MAG is going to own that business. I think that 10Gen and other companies providing baseline services are going to have a rough time playing that game.

Startups who wish to compete in the cloud business will have to provide great value added services that facilitate unique new application categories sitting on top of one or more of the MAG clouds. The services will have to be hard to copy and/or narrow enough to not attract the attention of MAG.

Given the need to innovate in some unique way, and the need to be interoperable with MAG clouds, I am not at all clear how you can create innovate cloud platform services using an open source business model in a money making way. Being open source in this space is akin to what it might be like if Apple made OSX open source and optimized it to run on standard Intel PCs. Good karma perhaps. Good profits, not so much.

Of course many open source businesses hang their hat on services, consulting and support. I personally hate time and materials type businesses masquerading as scalable software businesses, but my opinions aside, these are by and large tough businesses to succeed at.

In short, while being open source may be politically correct, I fear it may be a grave hindrance towards providing a defensible, unique, money-making offering in the cloud.

1 comment:

  1. The Cloud technology used by the provider is not interesting to buyers - the programming platform is. Right now it's about Java/.net/LAMP etc., and later we can probably put a full virtual appliance online on a cloud.

    ReplyDelete