Joel Spolsky has a great article this week about the Architecture Astronauts at Microsoft. I have to agree with core tenet of his post. In my experience, the most successful products are written with a particular end user application in mind. If you start with infrastructure, the risk of being wrong is huge.
I think that's the reason that Amazon's Web Services have been so successful. They've taken existing capability and made them better rather than inventing a whole new world.
1. EC2 - Amazon's virtual server hosting product, basically took an old service (dedicated server hosting) and made it better by automating it with a programmable API and lowering the cost substantially. However, the key was that they kept most of the same core features and functionality that companies had come to expect in their hosting providers. 2. S3 - Amazon's storage service took online file storage (done by many companies previously) and made it cheaper and more automated. Again, not creating a revolutionary new service, but just making something that existed already better.
Microsoft, with their new Live Mesh, is making a step to provide all kinds of new infrastructure that nobody asked for. Maybe they will be successful eventually, but I think it's a product looking for a customer.
My suggestion... start with a product that meets a real customer need and then build infrastructure around that product that makes sense. Starting with infrastructure and then looking for product applications is a risky strategy and doomed in the long run.
Build Products, Not Possibility
Joel Spolsky has a great article this week about the Architecture Astronauts at Microsoft. I have to agree with core tenet of his post. In my experience, the most successful products are written with a particular end user application in mind. If you start with infrastructure, the risk of being wrong is huge.
I think that's the reason that Amazon's Web Services have been so successful. They've taken existing capability and made them better rather than inventing a whole new world.
1. EC2 - Amazon's virtual server hosting product, basically took an old service (dedicated server hosting) and made it better by automating it with a programmable API and lowering the cost substantially. However, the key was that they kept most of the same core features and functionality that companies had come to expect in their hosting providers.
2. S3 - Amazon's storage service took online file storage (done by many companies previously) and made it cheaper and more automated. Again, not creating a revolutionary new service, but just making something that existed already better.
Microsoft, with their new Live Mesh, is making a step to provide all kinds of new infrastructure that nobody asked for. Maybe they will be successful eventually, but I think it's a product looking for a customer.
My suggestion... start with a product that meets a real customer need and then build infrastructure around that product that makes sense. Starting with infrastructure and then looking for product applications is a risky strategy and doomed in the long run.
Posted at 08:19 AM in Commentary | Permalink