Tom Rosamilia, General Manager System z introduced Karl Freund, VP z marketing who revealed the zEnterprises key features:-
- A “System of Systems,” integrating IBM’s leading technologies (System z, System x and POWER systems)
- The world’s fastest and most scalable enterprise system with unrivalled reliability, security, and manageability.
- The industry’s most efficient platform for large scale data center simplification and consolidation.
- The new IBM zEnterprise™ 196 (z196) for traditional z workloads and the familiar Linux on z operating systems, plus a water-cooled option for energy efficiency
- New zEnterprise BladeCenter Extension (zBX) to run POWER7™ and System x® Blades for tens of thousands of AIX® and Linux applications
- zEnterprise Unified Resource Manager, part of the IBM System Director
family, extends System z qualities of service across workloads
The stats are impressive, 50 BIPS, 3TB of memory, internal bandwidth of 288GB/sec (that's Gigabytes per second folks!) giving an ideal server for heterogeneous workloads, therefore perfect for data center consolidation or a private cloud.
That's the news, and you can read more all over the web, here, here and here for example, and here you can find the IBM announcement page here, for more information. But today I'm just going to look at a couple of things. Firstly the Security implications (as this is the zSecurity blog). And then I'll ask: "does this change everything"?
zEnterprise "Reduces risk by extending the reach of System z qualities of service". With a new Blade Center controlled by the System-Director-derived Unified Resource Manager and managed by the zEnterprise hardware console, your distributed systems should get a level of Resilience, Availability and Security approaching that enjoyed by System z for decades. The Unified Resource Manager will discover new components in the zBX and perform auto disaster recovery. And z/VM guests and blades will monitor themselves and log errors or call out an SSR.
Because your zBX is linked to the z196 by a short, enclosed, dedicated 10Gb/s network, there are no network cables to tap or sniff. And with capacity for 112 blades (which can be virtualised to many thousands of servers of course, plus you can have tens of thousands of Linux on z inside your z196, remember) you won't need servers all around your facility which need to be clustered, mirrored and locked down. Everything's in two boxes in your secure room being managed by the URM.
Data integrity is improved, with RAIM memory, yes RAID-style RAM with auto-correction of errors, and you can have CryptoExpress3 with up to 8 PCIe cards (like the z10EC) and some new SSL acceleration features, but of course URM makes all of this available to your AIX or Linux on System z blades too. So If your organisation runs highly available, highly secure applications on mixed architecture at the moment, then nothing else comes close to delivering zEnterprises levels of service.
Does this change everything?
It should. The old Mainframe versus Distributed war of words looks a lot like Microsoft versus Apple from a distance. Both camps have supporters with strong, entrenched, almost religious devotion to the platform and resistance to the alternative viewpoint. Some highly-paid analysts' ignorance of the recent development of System z borders on negligence, as I have blogged before. But this time feels different. Partly because this is no ordinary product upgrade. zEnterprise was developed over 3 years, costing USD5bn, and performed in concert with several large clients in Insurance, Retail, Finance and Healthcare, and Citigroup are an early adopter
And mostly because the obligation to choose between System z and Distributed architecture - between z* and Unix-like operating systems, between a monolithic mainframe and a collection of distinct, disparate systems each designed for a specific purpose and fulfilling a business need - has been eliminated at a stroke. Get zEnterprise and you get the lot, whatever you need. Traditional objections fall away, including the cost of porting apps to z from native *nix systems and the cost of z skills.
Joe Clabby was the 3rd speaker yesterday, an analyst and mainframe supporter (zPhile?) from Clabby Analytics and he suggested that the zEnterprise can jump you in front of your competitors by "at least five years". That's quite a statement, but clearly zEnterprise is not your father's mainframe. I've been suggesting for a while that we should perhaps stop using the term mainframe, and I tweeted a question a few weeks back asking for alternative terms for the latest System z machines. "Super Server" was suggested. But I think in time the generic term will become "Enterprise Server".