I think ec2 is the counterexample here. EC2 costs vastly more than owning hardware, and is in many ways harder (you have to design your application to be able to lose any server /much/ earlier than you do with hardware backed by a good SysAdmin's pager,) but from what I've seen working at large corporations, it's far easier (and to further contradict what I said earlier, cheaper) than running those servers in-house, at least for companies that aren't very IT focused. It's also vastly faster; as soon as the budget is approved and you have a guy trained on bringing up ec2 stuff, bang, you have your hardware, whereas in many companies, you ask IT and it's a long back and forth while IT tries to "help" you figure out what you really want, then it's another week while they get the hardware, then it's not just what you wanted, so there's another back and forth of a week. If the hardware was of a new type, there's another week while weirdness due to the new hardware are worked out. During all this time, the manager that wanted the servers can't do anything but whine to IT. With ec2, that manager can just get a better ec2 sysadmin and get on with it.
Therefore, many big company managers just head to amazon for their hardware rather than to the IT department. Part of the efficiency of ec2 is that you don't have to go through some salesrep and spend weeks going back and forth on price like you do when you buy pre-built servers (or way worse, data centre space or bandwidth.) There is no "vendor relationship" to manage, and "vendor relationships" add rather a lot of hidden costs on both sides.
My point is just that if you can solve the problem in a way that managers can plausibly justify an end-run around existing obstacles? I mean, you do need hype, and name recognition, but you can do it without a whole lot of high touch sales and "customer relationships."
Comments
I think ec2 is the counterexample here. EC2 costs vastly more than owning hardware, and is in many ways harder (you have to design your application to be able to lose any server /much/ earlier than you do with hardware backed by a good SysAdmin's pager,) but from what I've seen working at large corporations, it's far easier (and to further contradict what I said earlier, cheaper) than running those servers in-house, at least for companies that aren't very IT focused. It's also vastly faster; as soon as the budget is approved and you have a guy trained on bringing up ec2 stuff, bang, you have your hardware, whereas in many companies, you ask IT and it's a long back and forth while IT tries to "help" you figure out what you really want, then it's another week while they get the hardware, then it's not just what you wanted, so there's another back and forth of a week. If the hardware was of a new type, there's another week while weirdness due to the new hardware are worked out. During all this time, the manager that wanted the servers can't do anything but whine to IT. With ec2, that manager can just get a better ec2 sysadmin and get on with it.
Therefore, many big company managers just head to amazon for their hardware rather than to the IT department. Part of the efficiency of ec2 is that you don't have to go through some salesrep and spend weeks going back and forth on price like you do when you buy pre-built servers (or way worse, data centre space or bandwidth.) There is no "vendor relationship" to manage, and "vendor relationships" add rather a lot of hidden costs on both sides.
My point is just that if you can solve the problem in a way that managers can plausibly justify an end-run around existing obstacles? I mean, you do need hype, and name recognition, but you can do it without a whole lot of high touch sales and "customer relationships."