Skip navigation

Oh, this is going to be a good one. I’m going to let loose one of the biggest, meanest bees in my very full bonnet. This is one of the things that makes me Always Grumpy, but you know the old saying: the bigger they are, the harder they fall! This won’t take long to describe so it will be a shortish post.

If you’ve been reading my blog you know that I work in I/T. I work in an “enterprise shop”, which is some people’s way of saying that the corporation I work for has a large I/T footprint. It’s true! And like most large I/T shops we have many kinds of CPU, from big, blue mainframes down to servers under desks. We also have many kinds of network switch and router, but we won’t get into that here.

Servers are often categorised into three groups:

1)      Mainframe Systems are large systems that share resources (CPU, memory and I/O) among relatively few system images that each run a large number of heterogeneous workloads. IBM’s System z machines running the z/OS operating system are a common example, but there are others.

2)      Mid-range Systems are usually somewhat smaller in terms of overall capacity and share resources among relatively few system images that each run a smaller number of heterogeneous workloads. For example, many shops use HP Non-Stop machines this way.

3)      Distributed Systems are characterized by operating systems that run best with only one or a few distinct workloads per system image. So there are often many distinct machines each with its own system image or, more recently, fewer distinct machines sharing resources among many system images. These machines are often rack-mounted as “server blades”. Most shops have at least some machines running Microsoft Windows or various distributions of Unix or Linux as distributed systems.

So where’s the bee? I’m Often Dopey but have I completely lost my train of thought in my own thatch of mixed metaphors? And where are those Open Systems anyway?

Over twenty years ago, I/T industry pundits and manufacturers and re-sellers of Distributed Systems announced the “death of the mainframe.” Ten years passed and the dinosaur didn’t die. The millennium came, y2k passed and still the big blue beast and its cousins hadn’t given up the ghost. But Distributed Systems had proliferated: Windows servers popped up like mushrooms under desks everywhere, then assorted flavours of Unix started to turn up too.

Around that time, people started talking about Open Systems. Here comes the bee!

In Open Systems, the software and hardware were not tied to a single vendor, so the user’s enterprise wasn’t locked in. Windows was never an open system in any sense of the word. Unix was originally an open system: anyone could (and still can) get the kernel and a lot of other system software for free and modify it at will. You could (and still can) also get open source middleware and applications. But enterprises like their systems stable and supported, so they bought their Unix systems and all their software from vendors who would lock down versions to make them stable and sell licenses and support agreements for those versions. In other words, they’d make and sell a non-Open Unix system. The same thing quickly happened with Linux since its more recent introduction.

Wait, where’s that bee gone? This all sounds perfectly reasonable!

Here’s the bee: lots of people refer to Distributed Systems as Open Systems. I hate it! But why, you ask? Because “Open Systems” sounds so tasty and so nice, so friendly and approachable. “Open Systems” sounds a lot better than icky, lame old mainframes and mid-range systems, which by exclusion must be “closed”. Distributed Systems are called “Open Systems” by people who are not being careful about their words: they are delivering a hidden value proposition every time they use the phrase. I won’t say that everyone who does this is malevolent – I’m such a nice guy! – but many of the people who use the incorrect phrase have a vested interest in making distributed systems look better than mainframe and mid-range systems. They talk about the hundreds or thousands of “Open Systems” servers they run in their enterprise I/T shops when they mean Windows and stabilised, supported versions of Unix and Linux. Most of them know better and many also know they’re mis-representing: the truth is that enterprise I/T shops rarely run systems that are in any way open. In enterprise I/T shops we usually need high availability, stability, security, homogeneity and managed processes to satisfy our customers’ requirements; we sacrifice agility and low cost to achieve those goals. To do that, we install stable, supported hardware and software. In other words, enterprise I/T shops almost never run any Open Systems and always run many Distributed Systems.

There, I’ve freed the bee! If you want to get rid of the bee, never refer to your distributed systems as “open” unless you’re running a free, unsupported, open source distro OS on unsupported and modifiable hardware with free, unsupported, open source applications. Then they’re open!

Oops! That wasn’t a shortish post, was it?

alwaysgrumpy is Jonathan Gladstone,  or follow me on Twitter @jbglad59.


Go ahead and comment... don't be Bashful!

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: