Last week I elaborated on the first value in the "what" dimension of the Mainframe Analytics Taxonomy.
Before I dig into the second value, however, I have some good news: our official website is finally up and running at http://MainframeAnalytics.com. Feel free to check it out and offer any feedback.
Now, concerning interfacing with devices, of course it is primarily the operating system's role to enable applications to talk with the terminals, tape and disk (and solid state) drives, printers, network, etc. After all, that's one of the main jobs of operating systems: handling the stuff that every application needs but isn't the core functionality that the application is about providing.
However, it's one thing to get data to and from these devices. It's entirely another thing to manage and optimize the usage of these devices.
For example, network problems can occur anywhere between the mainframe and the user, and tracking them down and fixing them can be nearly impossible without network management software. In fact, such software can often prevent or detect such problems before anyone else is affected.
Sharing drives and consoles between multiple different operating system images is also a challenge - for example, making sure that a change made to a drive on one system doesn't accidentally overwrite a change made to the same place on another one.
The software that manages and optimizes these devices is a core part of what makes the mainframe great. While storage, network and other device management software may be available for other platforms, it is not yet nearly common enough that the time and expense are committed to do this as well as is standard on the mainframe.
This is a common enough thread in understanding the value and role of the mainframe that it bears emphasizing: while many of the things that make the mainframe great may also be available for other platforms, it is generally the exception that they are purchased, installed, configured and run properly on those other platforms, but normal to have them on the mainframe. In fact, if a non-mainframe platform were run with all of the things that make a mainframe great, and if all these additional systems didn't degrade that platform beyond usability, it would still make the cost-benefits equation favor the mainframe by a very large margin.
It may be tempting to think of the mainframe as expensive, but for the quality of service, availability, security and reliability that we expect - and which keep the world economy functional - the price we pay is very small indeed, especially compared to the costs of the consequences of doing without.