bigger and bigger...

I checked again and we're now at an estimated 78,500 pages talking about SceneCaster today according to Google. The good news for me (I'm expected to make sure that our service is up and operational 24/7) is that we're not doubling every day any more. From 54K to 78.5K in one day yields a daily growth rate of 1.453. Doing the math once more (can you tell I was trained as a mathematician?) with this rate of growth I have roughly 20 days ahead of me, instead of the 10 I was guesstimating yesterday, before we hit 50 million references to SceneCaster on the web. I still need to throw a heck of a lot of machines into rotation to make sure that we handle the associated growth in load appropriately, but so far we've handled the spike very well and our servers are handling the current growth without a glitch

My nightmare scenario going into this launch was that we might not be able to meet the load requirements. My nightmare scenario was definitely brought to light just before our launch. Mint.com was awarded the techcrunch top company award a week before we launched. As a result of the interest in their product their site went down under the barrage of requests.The last thing I wanted to see was a landrush of new users being able to choke our servers with their requests. We took precautionary measures using Akamai's EdgeSuite to help us absorb most of the static data serving we might get hit with and focused most of our effort on scaling the dynamic aspect of our service. That included doubling the number of database servers we expected to need for our projected growth, just in case. I look at our statistics right now and can see we were right in doing so. The load is heavy, but manageable. No outages to date.

The interesting thing that I note as I look at the growth patterns is that almost everyone jumping on the SceneCaster bandwagon right now is doing so via our main web site rather than our facebook app. The ratio of new users is about 10 to 1 main site versus facebook app. What this tells me is that the majority of new users are coming in due to the press releases (which all mention the URL for the main site and never explicitly indicate the URL for the facebook app). What I'm watching for is when that trend inverses. Once people start using it, I would expect that the major vector for new users to be existing users and the best way for that to happen is on facebook.

If we were to do this again, and this is a suggestion for those of you out there thinking to launch a service that has a facebook app component to it: publish the facebook app URL explicitly in all communications about the product. Facebook users can then directly hook in to your solution from facebook itself. It makes the startup cost minimal and lets you rely on the work that the folks at facebook have already done to validate that each user is a real human being as opposed to a spambot, so you don't have to do that validation yourself.