Since the 1990s, I guess, whenever I mention that I have some connection with mainframes, I feel as though I am on the defensive. It’s as though the general zeitgeist is, and has been for the past 30 years, that mainframes are old fashioned and barely hanging on in there, in a world that has moved on.
My argument against that has always been to look at aeroplanes and cars. If you want to move lots of people quickly from one place to another, then you need a large passenger plane. If you want to move just a few people you’d use a small passenger plane. There are plenty of different makes and models you can use to illustrate this. And, of course, there are all sorts of makes and models in between these two sizes, and there are things like hang gliders and gliders that also have a role. So, what the aeroplane metaphor shows is that there is a need for different types of aeroplane for different needs and uses. There’s not a one-size-fits-all approach to planes, and similarly, there’s not a one-size-fits-all approach to computing. Cloud is great, but no-one is getting rid of their laptop, or tablet, or even their phone, so why should they get rid of their mainframe?
And, if people like cars, you can use a similar metaphor. There are plenty of big cars being sold as well as Teslas and other electric cars. Different people have different needs. Some need a car that can fit seven people and their luggage in, some need a car that can carry Ikea furniture home in the back or take rubbish to the recycling centre. Some want a really small car that is easy to park. We don’t all drive around in identical vehicles. And new cars are all very different, in so many ways, from cars in the 1960s.
But let’s suppose that there never were any mainframes. This is a parallel world where they just weren’t produced. That’s not too big a stretch of the imagination because it was very expensive to create the System/360, and quite difficult to get the original operating system (OS/360) working on it. That’s why DOS/360, and even BOS/360 and TOS/360 saw the light of day.
In this non-mainframe-world scenario, you would still have the client/server environment of the 1990s, and you would have the current cloud environment, but what you wouldn’t have is a highly secure, centralized location where batch work would be able to access data very quickly. There would be no issues of losing computing power because the Internet had gone down.
So, that world would have to invent the mainframe now! And, although the hardware would most likely come from IBM, and the operating system definitely would, the applications that run on the mainframe could be sourced from any number of vendors. In addition, peripheral devices could come from a number of different vendors. It would be the increase in the speed of batch jobs that would be the most noticeable improvement following the invention of the mainframe.
There would also be huge cost savings – yes, I did say huge cost savings – for sites that were used to having large server rooms running multiple Linux servers. Each of these would probably need its own team of people ensuring everything was running smoothly so that end users could continue working. Our newly-invented mainframe would be able to run Linux with far fewer support staff and even more end users. So, firstly, there’s the saving in staff costs, but also, a much smaller server room would be required, and that would require purchasing less hardware, using less cooling, etc etc. Organizations would be jumping at the chance to get on board. And, of course, a complete mainframe can be rack-mounted in the same way as other servers. So, there will be space for it.
While CICS and IMS are both brilliant subsystems on mainframes, IMS might well be seen a as a dream solution for financial organizations that needed to process financial requests as quickly as possible. The various IMS database access methods have always made this speed of access a central part of the way they work.
When it comes to security, mainframes, without doubt, do it better than other platforms. Firstly, there’s never any issue with groups within a large organization going on and buying cloud services without the IT team knowing. There will be no shadow IT because everything is going through the mainframe. The mainframe ensures that people can only access the applications and data that they are meant to access – and not everything that’s on the mainframe. It can also ensure that data is automatically backed up. It can ensure that an audit trail is kept of what happen to data, when, and by whom. This will ensure that an organization is compliant with whatever regulations apply to it. In fact, a mainframe is one of the best platforms to move to a zero-trust way of working.
Mainframes are reliable, available, and serviceable. They have more than on processor, and more than one logical partition. Multiple machines can be clustered in a parallel sysplex. They can connect to other platforms (like cloud, mobile, etc). They can be highly automated, and they now come with software making it easier for people without much previous mainframe experience to work on them and control them.
What I’m arguing is that we should reframe the debate about mainframes being your dad’s technology to one where we argue that if they didn’t exist, we would have to invent them.
No comments:
Post a Comment