by Bob Walsh
Last week, fairly quietly, Amazon changed what computing meant. It’s almost as if they snuck into every home, office and company, took a little white-out to the dictionary, and completely disrupted what servers used to mean. It will take a while before people notice the change, but trust me they will.
Last week Amazon announced EC2 – Amazon Elastic Compute Cloud – as a limited beta. For $.10 an hour, you can programmatically create as many servers as you need on Amazon’s heavy iron. For $.20 a gigabyte your servers can talk with Internet. And for $.15 a gigabyte your Amazon virtual servers can store and retrieve data from your Amazon virtual hard disks (see Amazon S3).
So why does this truly change the rules of the game? How many web 2.0 start ups have choked and crashed their first day because the spotlight of Internet attention fried their servers to a crisp? How many problems do companies have that require massive computing resources to solve them, but only for a day? If I can run a server with good specs, why did I just buy a new Dell 1800 PowerEdge? Because I’d rather pay Dell $150 a month instead of Amazon $72 a month? Not!
For micro-ISVs, EC2 has the potential to make it possible to create programmatically all sorts of applications that used to be the domain of middle to large sized ISVs and VARs. Further, it gives us the ability to scale at will, and in line with our micro-ISV revenue.
Here are couple scenarios:
- You know a particular industry extremely well. And in that industry every so often there arises a problem that requires massive computing power — but only for a very short period of time. Maybe it’s the world of pharmaceuticals. Or maybe it’s video and animation. It’s probably something that nobody else here has ever heard of! You build an application that creates, provisions, starts and tear down the computing processing power your customers need at EC2. They pay EC2 – you supply an micro-ISV VMWare Workstation for them if you will.
- You decide to build an app designed to handle massive amounts of video streaming for relatively short period of time. Think of the Olympics. Or think Little League baseball all over the country when parents want to see their kids at bat. You design the virtual machine (Amazon calls it an Amazon Machine Image) that as an instance or in clusters can handle the demand.
- You design a virtual fallover application for all those actual ISPs and software as a service companies out there who one time or another fall down. For example, stikipad.com would not be losing me as a customer today (for the wiki I’ve been planning for this site) after their wiki service crashed and burned this weekend if they’d had already bought your micro-ISV virtual fallover application.
Of course, this is all still in beta, and limited beta that. You you can instance Linux servers, but hell will freeze over in my opinion before Microsoft lets you instance it’s servers. Or perhaps not — Microsoft today doesn’t think the way Microsoft used to think.
Not being at heart a server guy, I’m sure there are many other ways that being able to instantly and programmatically instance as many servers as you want, with what you want, on a reliable platform can be used to turn a buck. I do know that for my own micro-ISV plans, Amazon’s EC2 is going to be something I will be keeping a very close eye on. If this pans out, it makes it makes it possible and economical to build a web app I’ve had on my brain for over a decade. Time, I hope, will tell.
You can read more about EC2 at this blog, this blog, this blog and this blog. These guys know their server stuff better than I ever will, and if they are excited, maybe you should be too.
[tags]E2c, Amazon, servers[/tags]