I originally wrote this down as an idea for a blog post back in 2008, but I never got to it. Last week I heard someone talking about how computing would move closer to data since computing power is cheap and moving data around is still expensive.
Storing large amounts of data locks customers in since moving data around is still slow. Part of the reason is that Moore’s Law tells us computing power will double in 18 months, but it takes longer to upgrade the infrastructure needed to move data around. If a company, like Amazon for instance, houses a customers data they can easily deploy more processors near that data. This makes operating on the data at Amazon more attractive, while the cost to move that data to a new data center drops at a slower rate.
For SaaS and cloud companies, having large amounts of their customer’s data stored there is a natural barrier to entry to competitors.