Linux on System z

Should IBM drop the “Mainframe” Moniker?

IBM announced its latest Mainframe system yesterday which includes even more support for Linux and Cloud. This morning, I came across this article in which Quocirca's founder Clive Longbottom suggests that companies are put off from considering System z as a valid Cloud platform because they associate the mainframe with a monolithic, outdated technology.

"Call it anything – call it Jennifer if you want – just get away from 'mainframe'. IBM has got to change the way it is messaging these items," Longbottom concluded.

So, I kind of agree…but with concerns. I’d love companies to see System z as the amazing platform that it really is and see that for many types of workload it really cannot be beaten. However I see two problems with changing the name and pretending that it really is just another platform:

1. We (the mainframe community) need to work hard to stop the reality of owning a mainframe from being so different from owning a mid-range server. If you have to staff up an entire new team to manage the platform then that’s going to be a blocking issue especially if those resources are not fungible. We need tools to run on the mainframe that are either the same tools as the mid-range or at least use a familiar paradigm. And no, simply putting a 3270 emulator onto an iPad is not the answer!
2. If we renamed System z as System c – “the Cloud System” then we might end up not actually winning in any space – the cloud community might be impressed (but they’ll expect to pay mid-range prices not mainframe prices) and those customers whose companies depend on the features of the mainframe might think that IBM has lost their way and be more easily convinced to move away from the platform.

The solution? Maybe we bifurcate the platform into the conventional zEnterprise Mainframe and the lower end, competitively priced, business class Cloud machines? Longbottom definitely has a point though…the people who buy Linux severs today are often turned off as soon as they hear the word “mainframe” and never get to understand the real value of the platform.

http://www.channelweb.co.uk/crn-uk/news/2284286/analysts-tell-ibm-to-ditch-mainframe-terminology

Minimum Viable Platform Selection

A while ago I wrote an article that suggested that the minimum number of platforms that many companies could hope to optimize onto would be two – their mid-range platform of choice and the mainframe. For the foreseeable future they are not going to get down to one common platform – that would mean either getting rid of the mainframe or moving the distributed workloads onto the mainframe. Oddly, the latter option is actually possible although the former isn’t for many companies.

I am spending 2½ days with some of our customers this week discussing product strategy and of course the cloud is a topic of great interest. I realized that if I was a gambling man and was asked to predict the most likely way that a company would get to one on-premise platform this would be my prediction:

Mainframe and off-premise cloud

 

 

 

Think about it – there are hardly any systems left that companies would not move to the cloud except those that contain their most valuable, sensitive and protected data. You know, the kind of content that the company could go bankrupt without, the stuff that needs to be available 100% of the time…sound familiar?

So the model is quite simple:

• Get your compensation liked to the savings that you can bring to IT
• Move anything of high value onto the mainframe
• Get the rest out of the datacenter and into the cloud
• Close your massively expensive data center
• Retire!

This might just sound like a Mainframe Bigot's over idealist view of the world and to be fair it probably is except companies are really pushing to close their in-house data centers and I heard someone quote that by 2020 over 90% of the IT spend will not be in centralized IT services. Maybe this is a model that if we just aimed to achieve we could make some significant savings without reducing service levels.