Ten years ago, many people in our industry were predicting that the mainframe would be effectively dead within a decade, killed off by the rise of distributed computing, super-powerful PCs and newer, apparently more exciting options such as the client server. Today, however, there are many indications - including some hard facts - that reports of the mainframe's death have been greatly exaggerated.
There are several reasons why this is so, and why banks, insurance companies, airlines, supermarkets and telcos - among other enterprises - have found the mainframe a hard - indeed impossible - act to follow. Reliability is of course high on the list, with many machines having a record of a decade or more of running without interruption, and with hardware upgrades taking place during normal operation. Manifestly that is a benefit that can never be overlooked in the many applications where downtime would be costly or even disastrous.
Availability is another factor, with the ‘five nines' - i.e. 99.999% availability - often being considered as the rock-bottom starting point, not the ultimate goal, in day-to-day commercial mainframe operations. On top of that, there are the unmatchable security features, the high inbuilt redundancy and disaster recovery, the extensive input-output facilities, and, last but not least, the sheer processing power and MIPS ratings that support the massive throughput often required by our larger enterprises.
All of which explains why the mainframe continues to cling onto its established ecological niches. But is there really now a prospect of more than that - of a mainframe revival, in fact, despite all the interest in distributed computing? Well, a glance at the websites of some of the main contenders, such as IBM, HP and Sun, shows that the battle for ‘big iron' is hotting up. Of course the terminology used varies, with what is in truth a mainframe (or at least a quasi-mainframe) often described as a ‘large enterprise server' or an ‘extreme-scale system' or whatnot, but the message is the same - the big beasts are fighting, and in many cases winning, a remarkable comeback. One reason is the current interest in cloud computing and virtualisation, with many people waking up to the fact that these concepts - often touted as brand-new - have been realities in the mainframe world for decades.
There is fighting talk from IBM, for example, which claims that a single System z10 mainframe is equal to nearly 1500 x86 servers, but uses up to 85 per cent less floor space and up to 85 per cent less energy - an important plus given today's energy prices. The company also highlights the increasing number of software developers homing in on its mainframe platform, with references to ‘more than 5,500 commercially available mainframe software applications from 1500 independent software vendors.' And in terms of commercial results, there are some highly respectable numbers for a platform once written off as ‘legacy'. According to IBM, worldwide System z revenue for 2008 grew 12 per cent with MIPS growth at 25 per cent; additionally, installed MIPS for Linux, Java and database specialty engines grew at 77 per cent, 61 per cent and 56 per cent respectively.




Be the first to comment on this article!