Many computing professionals over the age of 30 can remember the days of the mainframe at University. Systems that required large specialist air conditioned rooms, support technicians and systems that were, in comparison to the non-technical user base of today, only intended for those with specific requirements who had the skills to use the system. I particularly remember, having come from a non-computing background how user unfriendly the mainframe systems were considered to be.
The infrastructure required for local site mainframe systems was large with the use of (mainly) thin clients that relied upon the mainframe for the processing power and multi-site usage became even more demanding given that communication line speeds were extremely expensive and slow. However mainframes were reliable, could handle large amounts of data and were specifically designed for robustness and performance.
Inevitably, as components became mass produced, clients evolved into having their own processing power at a much reduced cost and, as an important consideration, they could connect to many systems independently (you had to log on to a mainframe to access data) that did not necessarily fall under a single mainframe allowing computing flexibility and the introduction of graphical user interfaces (GUI) that drove the easy-to-use PC-Based revolution.
Mainframes still exist today and operate successfully on large mission critical systems (e.g. SpeedServe who supply Amazon) and small single machine based multi-user systems, as defined by the operating system (e.g. Linux).
We are experiencing a shift back to a mix of mainframe and pc-based computing given that many systems (especially Microsoft and Google office software) are moving online. This is due to the planned future revenue streams of software. For example, why should you upgrade from Word 2007 to Word 2010 when the product features of Word 2007 are probably sufficient for the majority of users in the market? This is a future revenue problem for those who supply PC-based software, such as Microsoft, so they are moving to a more subscription based online system, similar to mainframe computing, to stabilise their revenue over future years.
It will be interesting to see what will happen over the next decade and whether revenue requirements will dictate that our PCs become thin clients, in some respects, and must logon to online mainframes to access our subscribed software.
Coronel, Morris & Rob (2009) Database Systems: Design, Implementation, and Management (9th Edition). Cengage Learning.
Rojas, P (et al) (2007) Mainframes Strengths vs. Client/Server Strengths [Online]. Available at http://www.personal.psu.edu/pwr5004/Mainframes%20Strengths%20vs.ppt (Accessed 16 May 2010).