IBM/360 series mainframe detail (credit: Pete Clouston)
“SaaS is like returning to the days when the mainframe ruled - a hosted application reached via terminals”
Fair assertion, right?
Nah…I’d contend SaaS (Software as a Service) and a mainframe are two quite distinct – and different – models.
Nowadays, SaaS and the modern mainframe co-exist, though distantly. The dumb terminals have gone for good as the modern mainframe has embraced more recent technologies such as virtualization – interestingly that being one of the economic enablers of Cloud infrastructure.
From hereon though, this post asserts there’s a lot to differentiate between SaaS and the ‘dumb terminal’ mainframe model of yesteryear.
Mainframe characteristic # 1: proprietary hardware and software
Now – open web standards aid connections to the Cloud, removing reliance upon particular software or hardware flavors, in places optionally augmented by open source code.
Then – proprietary hardware, including peripherals that were relatively unreliable by today’s standards – this was exacerbated by high cost vendor lock-in, lead time delays on replacement parts, the largely monopolised (again high cost) ’science’ (or was it ‘art’?) of software development, combined with general lack of maturity across the SDLC.
Mainframe characteristic # 2: niche processing uses – processing of financial transactions or census data for example
Now – SaaS is anything but niche in its solution spread, operating in both the personal (e.g. Pocketsmith) and business domains (e.g. Salesforce) . In places SaaS straddles both business and personal domains (e.g. Google Apps, Weebly).
Then – mainframes were restricted to niche, processing-intensive uses, requiring capital investment that would have been prohibitive for most companies. Mainframes tended to be used for ‘number crunching’ based upon predictable and well defined data structures.
Mainframe characteristic # 3: dumb terminal human-computer interface
Now – graphical, usable by visually impaired, millions of colours, hi-viz-fi (millions of pixels), device independent and ubiquitous, typically with substantial client-side processing oomph. As an aside: more on client side processing – note Google’s positioning of javascript performance as a viable differentiator between browsers, and the proliferation of Rich Internet Applications via technologies such as Adobe Air or Microsoft Silverlight.
Then – text based, mono-colour, low-viz-fi (sub-million pixels), proprietary, device-dependent
Mainframe characteristic # 4: scalable storage via magnetic tape drives and disk drum devices
Now – dynamically scalable up or down at the click of a mouse, with fair and highly granular billing reflecting actual usage
Then – data storage was scalable upwards at a not insignificant cost. For a customer that already had the means to outlay the capital for a mainframe, additional data storage, while expensive, was not going to be cost prohibitive. If you scaled downward in those days, you didn’t recoup part of your investment – not much of a market for costly second hand proprietary data storage systems.
So there you go; now’s an opportunity to form a conclusion, if you’ve not already done so…
