This week Amazon Web Services announced a new pricing tier for its Elastic Compute Cloud (EC2) service and in doing so has differentiated its offering even further. At first blush the free tier sounds like a free trial, which isn't anything new in cloud computing. True, the free tier is time-limited, but you get 12 months, and capacity limited, along multiple dimensions. But it's also a new pricing band. And for three of its services, SimpleDB, Simple Queueing Service (SQS), and Simple Notification Service (SNS) the free tier is indefinite. Look for Amazon to lift the 12 month limit on this service next October, because the free tier will drive revenues for AWS long term. Here's why:
A few weeks back I posted a story about how one of our clients has been turning cloud economics to their advantage by flipping the concept of capacity planning on its head. Their strategy was to concentrate not on how much capacity they would need when their application got hot, but on how they could reduce its capacity footprint when it wasn't. As small as they could get it, they couldn't shrink it to the point where they incurred no cost at all; they were left with at least a storage and a caching bill. Now with the free tier, they can achieve a no-cost footprint.
By creating the free tier, Amazon has essentially created a new incentive for thinking small. And free is sticky. When deciding which public cloud platform you want to be hosted on, how can you argue with one that costs you nothing when traffic is low?
Before you jump to the conclusion that the free tier is just a promotion and loss leader for Amazon and thus they will never make it permanent, let me point out that the free tier will actually make money for AWS. It does this in two ways. First, if you know you can shrink your application down to the free tier, you have less incentive to switch platforms and that means AWS can count on you as a customer and can count on revenue when you're busy. The free tier is teeny, tiny.
Second, Amazon is a master at the game of Tetris and this is the game of profitability in infrastructure -as-a-service (IaaS) and server virtualization. In the world of virtual hosting, there is a percent of sustained average utilization of the infrastructure where you cross over from red to black. At only 20% utilized, your virtual environment might be losing money as your on-going operating costs are higher than the revenue from hosting. As you ratchet up the utilization you hit a point where the revenue takes over. Let's call that 60%, as our research has shown this to be roughly the break-even point for an average IaaS environment. The number varies based on the efficiencies of your operations, the cost of your infrastructure and your hosting costs. Raise the utilization higher than this cross-over point and each new VM you host is pure profit. So your number one objective as an IaaS business is to keep the utilization above this line. If you are a traditional VM hosting business with 12-month contracts you don't have to worry about this because you set your VM hosting prices above this cross-over point, get customers to commit to a utilization level, assign the resources and work to fill up the next box.
In the IaaS business, you don't have long-term contracts so you need pricing that incents users onto the platform and forecasting models that ensure you build the right-sized environment for the expected demand. Get it wrong: lose money. Other that playing with pricing – because competition makes this difficult – how else can you play this game to win? This is where Tetris comes in. If you have a bunch of large and extra large VMs, you can fill up a box pretty quick but if they go away you gotta find replacements, otherwise you go negative on that system. Since the break-even point isn't a full system, it's one much less than full, you can hedge against the loss of an extra large instance going away by filling up the box with a bunch of smaller instances. And the more you incent customers to use small instances the more small blocks you have to fill up systems – taking you further above the cross-over point. And as we mentioned, everything above the cross-over line is pure profit so you aren't hurting profitability if some of the smalls are free, you're just impacting short term margins.
If you want to play this game to win you can bring three new types of pieces to play:
* Reserved instances – blocks (and revenue) you can count on long term.
* Small, cheap instances – and lots of them so you never have an empty hole in a row of the Tetris bucket.
* Spot instances – so you can fill boxes even higher and get paid to do this, then raise the prices (profits) and kick blocks out that are impeding reoptimization of the pool.
Funny how Amazon has just such pieces today. No wonder they continue to lead the IaaS market.
Are you taking advantage of cloud economics yet? If you aren't, you're running out of excuses.