IBM’s just launched its new zEnterprise server at a very interesting time for the market. So interesting in fact that’s its very hard indeed to sum up all the swirls and currents currently roiling enterprise IT. Cloud (public and private), consumerisation of IT, Big Data, devops, NoSQL and did I remember to say… cloud?
- IBM Software Group has just “acquired” IBM Systems and Technology Group. That is – the managers that have aggressively built SWG into a software portfolio sales machine are now being given the hardware reins. Or should that be reigns?
- Oracle is in the process of swallowing Sun, becoming a hardware vendor in the process. Don’t imagine that IBM’s reintegration strategy is a reaction to Oracle though- remember IBM has always been a systems company… its Oracle parking its tanks on IBM’s lawn, not the other way around. IBM is just moving forward with a grand reintegration which been going on since Lou Gerstner was in charge.
- Big Blue just had a surprisingly decent quarter in x86/x64 sales. z and POWER were soft, but mainframe revenues are always lumpy as customers wait for a major upgrade to arrive. I have been following mainframes since 1995 and twas ever thus…
- IBM is seemingly attempting something of a reverse Intel (making a transition from chip seller to memory supplier, following the money as the pendulum swings back) as it builds on Intel Nehalem foundation.
- IBM’s new eX5 range of Intel-based servers takes a “network attached memory” approach, decoupling memory from Intel’s motherboards using a memory bus called X5. Nehalem servers are already memory beasts, but Big Data means Big Memory.
So what about the zEnterprise?
Its a hugely powerful mainframe – 96 cores running at a eye water evaporating 5.2GHz. You will be able plug X Series and POWER blades into plex using -the IBM zEnterprise BladeCenter Extension, a logical extension to the central mainframe across other servers with management software designed to manage workloads across these systems.
5.2 GHz – damn- that’s the fastest chip ever. I can’t help thinking IBM really needs to start scaling out otherwise its a Hitachi Data Systems Skyline scenario all over again – but this time at IBM’s expense. Itanium has hardly been a huge success for HP.
You can’t argue with the performance though. 50 Billion Instructions Per Second (BIPS) – which translates into some faster workloads.
What is more I asked IBM Tom Rosamilia General Manager, Power and z Systems, IBM Systems and Technology Group, about heat generated and he told me the chips only run as “hot as an industrial toaster”. 96 industrial toasters in a small box? No wonder IBM is offering water-cooling. The energy characteristics actually look reasonable, but the toaster comment has been haunting me.
Seems to me the server losing its raison d’etre here is IBM’s System P line. Unix just becomes a personality, managed by the mainframe in the zEnterprise system of systems. Engineering will happen at the Rack and blade or zEnterprise level. That is not to say AIX won’t continue to make a decent app server plug in, but the new scale out workloads are going to Linux, the OS of the web.
Talking of Linux- IBM claims zEnterprise can now run 10k Linux images. Blimey.
In conclusion zEnterprise is a consolidation engine for IBM as much as its customers. It will bring a lot of things together, dealing with the industry trends I outlined in the intro. The system of systems story is certainly not new. IBM has pitched and repitched the mainframe as a security hub, a systems management hub, a data management hub etc.
What is different this time? The industry is waking up to mainframe advantages again. As EMC marketing CTO Chuck Hollis recently posted on his StorageZilla blog (utterly essential reading):
Indeed, EMC’s mainframe DNA impacts how we interpret terms like “non-disruptive” and “mission critical” and “high availability” and a lot else as well. I think that one of the reasons so many shops depend on EMC for their most important apps is that we understand serious IT — regardless of whether it’s z/OS, UNIX or an uber-large VMware farm on the floor.
There’s another important message here as well: the cool things we continue to do for x86/hypervisor stacks, we also intend to do in the z/OS world.
Because mainframes aren’t going away any time soon …
Couldn’t have said it better myself.
In case you’re wondering what the illustration is- answers on a postcard please. IBM is a RedMonk client.