If you are in any industry long enough, you see cycles: Things fall out of favor and then another generation brings them right back.
With automobiles, we started out with electric cars as one of the ways folks would get around at the beginning of the last century, but cheap gas did them in before they matured. Here we are 100 years later and the hottest car on the planet is a gorgeous GT out of Britain that supposedly has 700 horsepower, a 200-mile range, and 10-minute charging time. I want one of these things so badly I can taste it.
In tech cycles, this happens much more quickly. Back when I started, tech was all about mainframes, but they seemed to quickly fall out of favor. Even IBM was convinced that mainframes were dead. Here we are 20 years later, and they still represent one of IBM's most profitable businesses.
The reason is because cloud computing, by most measures, feels like a natural environment for a mainframe, and the concept could actually cause the platform to surge.
Back to the Future
What are the main advantages of a mainframe? Reliability, scalability, I/O and security. The disadvantages are that they are large, few folks know how to work on them, and they are nasty to replace.
In a cloud-based solution, reliability and scalability may trump everything else. We recently saw Google Apps and Gmail fail badly, and Apple continues to have major reliability problems with its MobleMe offering. While this is Apple's first real push into the cloud, it clearly had no clue what the loading and reliability factors were -- otherwise, it would never have released such a failure-prone product. On the Google side, here is a company whose middle name should be "cloud." Yet it failed as badly as Apple did.
Reliability, when you are talking about computing services as if they were a utility, needs to be absolute, and there are few who can provide this level of reliability. The two that come to mind are IBM with its mainframes and HP's old Tandem division, which it acquired through Compaq with solutions that eventually evolved into Itanium NonStop Computers. IBM calls its platform the much simpler System z.
While both companies pound on each other a great deal as to which solution is best, the reality is that both owe much of the continued and apparently increasing interest in mainframes to cloud computing and the need for a utility-like experience.
Could We Be One Step Closer to the Future Past?
Back when IBM was at its peak, it leased mainframes and was virtually recession-proof. Cloud computing, in its absolute sense, isn't computers you purchase for your own cloud, but services you subscribe to for your organization, which can be increased or decreased based on need.
With all that cloud computing brings to the table, I think we have the basis for a return to the past, in terms of lease-like services, but a new future where the charges can be adjusted to need. Better for both the vendor, because it isn't so sales-oriented, and better for the IT organization because the costs are less rigid than capital depreciation or standard leasing.
Cloud computing is ideally suited for large-scale, mainframe-like solutions, but increasingly provided as highly flexible subscription-based services. Let's see what the future brings.