Retired software engineer Billy Newport said that IBM mainframes were the first onsite private Cloud. Fancy computers and storage running custom software installed onsite.
Customers liked them because they offered a one-stop solution that lets them pay as they go. They had more than needed, and a call to IBM would unlock more power. IBM looked after the hardware and swapped parts when they broke down. IBM charged for service and storage. Pay as you go. Start cheap, and your costs will go up as you use more.
Newport said that it was a brilliant business model. Building apps and systems to run on this platform lowered costs and gave businesses flexible pricing as their business grew.
“The problem with it was any heavily used apps paid service prices based on their monthly use of services. As the app use grows, the bill grows and the control of the bill is mostly in IBM’s hands,” he said.
You use more, you pay more. If a company runs 100 apps on the mainframe then it’s bound to happen that most of the mainframe monthly cost will be a few of those apps. Apps almost always start small. The focus is on function and not on saving money. So, it ships fast and doesn’t cost much as it’s lightly used. But, a few apps become popular and are easy to grow. Sadly, while computers are flexible, budgets are not.
“Heavily used apps became costly to run. Even if you stop working on it, you are still paying money out the door to run it on the mainframe. Usually, developers were told to work on expensive apps as they grew to try and save money,” Newport said.
What did work well, I think, was that smaller computers (AS/400 or iSeries) and mainframes were places to host other companies’ apps. We call hosting other companies’ apps SaaS platforms today :) I think these were usually good values. The customer got a great set of functions running on a top-class kit for some helpful business purposes.
Newport thinks today’s Cloud looks precisely the same as the mainframe scenario.
If you step back and look at it. The previous bit also describes today’s Cloud and predicts the same problems. Companies have rushed to get on the Cloud with the cool kids.
“Many companies will try to rush to cut cloud spending and will find moving onsite to be a pricey option if it’s even possible,” he said.
“Cloud is mostly good for running SaaS, not for basic services. Think CRM/Payroll/development hosting/MS Office/Corporate email and so on, not data storage/kubernetes/virtual machine hosting. This is more valuable to a company than just hosting data or virtual machines,” he said.