Yawn. Been there, done that. In the mid 1970's, it was widely predicted that corporations would not own their own computers in the future. Instead, they would have 3270 terminals hooked up to hosted virtual machines on IBM mainframes via SNA, or they would have ASCII terminals hooked up to accounts on mainframes elsewhere via packet-switched networks. A few smaller companies did try this when they computerized their own operations, but ran into the following problems:
Service guarantees: If in-house, they can guarantee access to their data. If outsourced, they must rely on the vendor, and how can they trust the vendor?
Data ownership. Who owns the data -- the vendor upon whose computers the data resides, or the company? And what happens if the vendor goes out of business?
Security. How can they be absolutely sure that the vendor is not in cahoots with one of their competitors, or that someone else can't get access to their data?
Bandwidth. You have basically unlimited bandwidth to your computers locally, so your terminals are always super-responsive. The 3270 protocol with its half duplex screen-transmit-based paradigm was a work-around to that problem, but an imperfect one.
Data retention issues. Corporations have a legal duty to retain certain data for certain periods of time. They also have a strong desire to purge said data the moment it's not legally necessary to retain it, because if it's still around if (when) someone sues them, it can be subpoenaed and used against them in a court of law. Without ownership of the computing facilities, corporations have much difficulty enforcing data retention policies.
The end result was that when smaller minicomputers and microcomputers became available, the time sharing services went under in short order, because the small to mid-sized businesses whose accounting etc. was done via time sharing services swiftly brought that in-house.
Now I'm hearing the same arguments made about cloud computing -- how it's easier than managing your own infrastructure, about how much more reliable the large systems used for cloud computing are, etc. -- and from a business perspective, it's dejavu all over again. Perhaps individuals will embrace it for their personal data. But I don't see corporations putting any of their corporate data into the cloud, other than data that has to reside there anyway -- i.e., the customer-facing data on their web sites. And as long as corporations are reliant upon in-house computing systems accessed from personal computers, there will be a market for personal computers, if only to access that one application that simply doesn't work "in the cloud" (for me, it's the corporate source code control system). As for the notion that the fundamental structure of how companies do business will ever change: Err, no. We've been doing double-entry accounting for hundreds of years now. Computers didn't change the fundamental structure of how we do accounting, payroll, and other critical business functions of that nature. All it did was make it faster and easier. 'Nuff said on that.
]]>