I mentioned in my earlier post that it would be interesting to compare prices between spot prices and traditional instance prices. The data is already available at http://cloudexchange.org/charts/us-east-1.linux.m1.small.html and it looks like spot prices are waaayy lower than traditional prices.

Stay tuned and it is going to be interesting how this pans out.

My thoughts on cloud computing, lessons that we learnt at Pixily and my words of wisdom for young Jedi warriors are now available from xconomy. I hope to make some of my presentations available right from here in the near future

At the Xconomy Cloud^3 conference and unpanel last week, there was a lot of great discussion around the Cloud’s cost economics and at what point the Cloud becomes more expensive than a in-house data center. As part of the unpanel discussion, I had mentioned that cost models will continue to evolve in the cloud and that we are in the very early stages.

And here it is, the Amazon Ec2 Spot pricing. The concept is seemingly straightforward (with a somewhat unexpected twist). You can bid for “Spot Instances” where you specify the price you are willing to pay for an EC2 instance. To quote the AWS blog “As requests come in and unused capacity becomes available, we’ll evaluate the open bids for each Region and compute a new Spot Price for each instance type”. And here is the twist. After that we’ll terminate any Spot Instances with bids below the Spot Price, and launch instances for requests with bids higher than or at the new Spot Price. The instances will be billed at the then-current Spot Price regardless of the actual bid. I don’t know why the instances would have to be terminated, rather than a simple re-calculation of the price of running the instance, but this seems like a fun experiment.

To be able to fully take advantage of this kind of flexibility being offered, your application (actually I should background jobs) would have to be totally stateless and be able to resume from any where in a workflow. Queuing mechanisms such as SQS come in handy. Like I mentioned in my talk, it is probably good design anyway to be stateless. And of course, please don’t run your web server or user-facing applications in a Spot instance. That is a guaranteed way to get fired.

All in all, I think this is more of a pricing experiment from Amazon’s perspective, more than anything else. Getting every additional dollar out of your excess capacity is a centuries old problem (think free business cards from VistaPrint), and I’d be interested in knowing how this pans out for Amazon and how Spot prices compare to traditional prices.

I found this fascinating quote today:

I believe that in startups, like venture investing, the cost of making a bad decision is not nearly as great as the benefit of making a good one. So I like action oriented leaders.avc.com, A VC, Dec 2009

It’s cheating to start a blog post with a quote from Fred Wilson. He is that good. But sometimes you need to cheat and I’m doing it today.

I suggest that you read the whole article, which exactly captures what startups ought to be doing, and why the cost of not trying is way more than trying and failing.

I am a long time follower of the now-popular GTD methodology, and the recent blog post about Infusing GTD into Marketing makes perfect sense. With Marketing, as with defining features and developing them, it is important to define goals and know when something is “Done”. This, in a way, is a great corollary to “If you are not measuring, you are not marketing”. What better way to know if you are done, than when you measure. Here are a few examples of concrete goals that I can think of:

• We will stop the A/B test when we B outperforms A in conversion by half percentage point for atleast one week.

• The newsletter to convert free users to paid users will be tweaked until we see a X % improvement in conversion rate, Y% improvement in click through and Z% improvement in open rates.

This is especially difficult for startups, where resources are even more constrained than other companies. Asking what “done” and “success” look like are two quick and easy ways to infuse GTD into your marketing processes and realize some immediate benefits.

So, begin with the end in mind.