One virtual machine per fisherman: On IBM, Linux, Mainframe and the cloud

I sat down with some analysts from different firms for a group briefing this morning with Olivier Alluis, IBM Systems and Technology Group (STG) cloud and systems software sales leader for SW Europe. When Olivier asked what I hoped to get out of the session I simply replied: “not much”.

It might have sounded a touch arrogant, but what I meant was that I know the other analysts and their focus areas, and we all look at cloud differently.  Peter Thorne from Cabashi for example was specifically interested in how manufacturing companies are using cloud computing.

So I listened to Luigi talk about develop, test, deploy – which I tend to call the VMware pattern (please see coverage here), and about IBM’s Cloudburst private cloud appliance. So far so expected. But then he said something that stuck a hook in my cheek:

“We need to kill the notion that cloud means commodity. We have some great examples of System z in the cloud”

System z in the cloud? That’s the IBM mainframe in the cloud. Really?

“OK then tell us about one,”

I said, expecting Luigi to talk about the dreaded unnamed “customer in financial services”. But no.

Universita di Bari is developing an application with low barriers to entry, intended to bring in revenues at a time of higher education budget cuts across Europe.  The college has built an online fish market, with local fishing boats as customers. Every boat has its own Linux virtual machine running on IBM z/VM – a system designed to manage hundreds, or potentially thousands, of machine instances on one server.

My immediate assumption was that Universita di Bari was already a mainframe shop, and had used some spare capacity for the new app. But it turns out this was a new win for the mainframe, which had been purchased (at significant discount) specifically for its capability to provide a community environment.

Each fishing boat in the fleet has its own virtual machine, to store information about their latest catch. A back end application then matches the supply to local restaurants, so the fisherman can sell his catch before he even gets back to port. Buyers get fresher fish, and fresher menus. IBM has built a similar system in Galway Bay, which is well worth checking out. The Bari app was successful enough that the university has also now made an equivalent for local wine makers and buyers.

All I can say is damn good work Luigi – your team deserves a bonus. Why? Well its like this- in my 15 years watching the mainframe IBM has always had an ambivalent attitude about winning new customers. Generally mainframe revenues increase as major customers buy more capacity, in IBM parlance MIPS. IBM doesn’t mind losing midrange customers to other platforms (hopefully its own System i or system x) as long as it keeps growing at the high end. A reductionist strategy perhaps but extremely profitable. IBM is generally happiest when it is selling to huge IT shops in the Fortune 500.

But here is a net new win, a small capacity box in mainframe terms, but a big win in marketing and impact, and obviously a pretty major commitment from a university. In order to sustain the mainframe going forward IBM has to refresh the skills base- pushing Mainframe = Youth at every opportunity. Bari is a canonical example.

You might push back and say this is just SaaS, rather than cloud, but that’s unfair. If we’re talking about spinning up linux instances on the fly, in completely automated fashion, with billing functionality built in – well that smells like cloud to me.

Obviously one win isn’t a trend, but this feels like a great proof point. I should stress that IBM is a customer, and that I have not talked to Bari directly. But at a time when VMWare is making great running in both public and private cloud, and the likes of Amazon Web Services continue to dominate the market, its pretty impressive to see IBM mainframe slicing as dicing as part of the cloud mix.

updates: IBM, I should have said, is a client. I am also extremely impressed by IBM’s latest financial results. The mainframe just had its best quarter in six years!