I was having lunch the other day with the CTO of a successful web startup which is on its way to financial break-even, usually a good sign that the business model is working and the venture capitalists will support the company until it goes public.  He asked me about ENKI's Computing Utility offering and why it would be advantageous for a company like his to use it.

Ultimately, for a start-up company living off of investor cash, the name of the game is to be able to break even with the smallest investment possible, which allows the founders, investors, and employees to make the largest profit.  Even with a great business model and happy customers, this means controlling costs.  In the late nineties, wild parties, huge speculative marketing experiments, and vast, optimistic computing infrastructure build-outs were the rage - but most of those companies ran out of money, contributing to the dot-bomb crash.   Working in that world, I always wondered what happened to business fundamentals: careful investments and cost control.  Nowadays, these best practices are back in style even in the high-flying world of web startups, in large part because investors are more careful about how they spend their money.

This is where utility computing comes in.  Instead of investing in your own data center and staff to maintain it, you just buy as much computing as you need.   Large enterprises talk about depreciation expense, taxes, and capital cost as part of their break-even financial models, but for the startup it boils down to "how much money is in the bank and how long can I run with that?"  So buying computing as a service makes a lot of sense.  But how much sense does it make?

Recently I was at the web2.0 conference and attended a talk by one of the founders of Digg, who shared a little about how much computing resource they were applying to their business.  After getting a rough idea of what they were doing, I asked myself how much it was costing them, and if buying utility computing from ENKI would have helped them out.  A little time with Excel and some numbers I put together from published colocation costs got me a spreadsheet that said that a Digg-like company would save about a million dollars over a year and a half on data center costs.

The spreadsheet can be viewed here .

I had to make some assumptions, particularly that the start-up was trying to run very lean, working their IT staff hard, in my experience to the edge of burnout.  This seemed in line with Digg's staffing count and my experience with web start-ups.  I also had to make some assumptions about how the startup's application would map into ENKI's grid technology, which allows for much higher levels of server utilization and failover capability with less hardware.  So the numbers might be off by 20-30%, especially depending on the startup's IT philosophies.  But no matter how you cut it, paying as you go is what a start-up should do.  

Another thing I discussed with my CTO friend was if a start-up could outsource their computing needs with confidence.  In my experience working with start-ups, their computing infrastructure succeeded or failed based on the quality of their IT staff and management.   At ENKI, we have an experienced team, but we also make the interface with our customers easy by not just selling hosting, but rather consulting with our customers to implement their applications successfully on our computing resources.  I accounted for this cost as best I could as a per-server monthly labor charge.  It is surprisingly low, but that's the benefit of the grid technology we're using.