Real Time View

HPE unveils next-gen Nonstop Compute
for mission-critical work

Safe to say, Nonstop Compute is safe. By this I mean, with the introduction of the HPE Nonstop Compute NS9 X5 and NS5 X5 converged systems, HPE demonstrates support of Nonstop. What it also signifies is that the Nonstop community voice was listened to in a manner that changed the course of the Nonstop product portfolio. Whereas the initial inclination was to pursue the creation of an advanced virtual Nonstop, polling of customer requirements identified little need for a virtual Nonstop and the decision to remain steadfastly behind an updated converged system was an easy decision to make.

On a personal level, this is something that didn’t come as a surprise. Closing in on six decades in tech you can readily see that I have seen it all. Perhaps not, but the general idea is not too hard to grasp. Talking of grasp, did you know that for the IBM mainframes of the late ‘60s and early ‘70s, there wasn’t an IBM supplied spooler. Basic, you might think, but no, this breakthrough mainframe that changed the course of computing, as a primary batch process had no spooler. It was left to an Australian, Boyd Munro, working for IBM at the time, to develop GRASP. Sold initially in Australia, then Britain before entering the US marketplace, many argue that it was the original independent software product ever produced.

Back in those times, I stumbled into the IBM software industry when I joined The Computer Software Company (TCSC) out of Richmond, Virginia, that developed an alternate IBM mainframe operating system, EDOS that came with an integrated spooler. It sold well for a decade or two even as IBM was providing its operating system for free. It wasn’t until the arrival of the IBM 4300 mainframes where IBM finally began charging for the operating system, did the interest in EDOS (and EDOS/VS) decline. Fortunately, in stepped Nixdorf Computers, bought TCSC, and relaunched EDOS as NIDOS just as Nixdorf entered the Plug-Compatible Mainframe (PCM) business.

All of which is to say, for almost my entire career, I have been involved with the independent software business almost from the very beginning. Secure in the notion that there really isn’t anything new under the sun, the decision by HPE to support another round of converged systems was not only a wise choice to make, it was the only decision to be made that made sense. Nonstop Compute is protected and as such, its future is safe even as the Nonstop community is secure in the knowledge that Nonstop lives for at least another decade – in its current form, that is, riding the ProLiant Intel x86 roadmap.

It continues to surprise me just how oblivious the overall IT industry is about the strengths of Nonstop Compute converged systems. HPE may not lead with Nonstop yet if the prospective client asks about Nonstop, then HPE is only too willing to step in and support any subsequent prospective client proposals featuring Nonstop. And why? It’s simply hard to argue with a product that doesn’t fail. Rest easy, CIOs, there will be no midnight calls informing you that yes, the system is down. It has crashed! Availability remains a powerful story favoring Nonstop.

It is against that background that I have been in a series of exchanges with my IBM colleagues. The most recent news from IBM marketing happens to be based on the new 9 Nines of availability. A reference to just a few seconds of downtime per year. But here’s the catch. This many nines is in reference to unplanned downtime. It seems that planned downtime no longer is considered downtime at all and be that as it may, it’s a serious omission as many of the IBM mainframe subsystems require maintenance to be routinely performed and that typically needs to be planned in advance as “planned system downtime” is required. Ouch!

The longstanding traditional IBM mainframe falls a longways short of matching Nonstop levels of availability. Certainly, for the adventurous IBM systems programmers, you can always move up to Parallel Sysplex, a mainframe clustering technology via a Coupling Facility, where the utilization of a minimum of two autonomous systems that share clocks, locks and lists (perhaps, a little more these days as it’s been a while since I last looked at Parallel Sysplex), whereby you can program in substantially higher levels of availability including minimizing planned downtime. However, the very need for the presence of gifted and adventurous IBM systems programmers to implement Parallel Sysplex leaves any IBM customer vulnerable should key members of the team leave the IBM customer. It’s just very hard to maintain across system generations.

I keep posting about this on social media and I am pleased to report that the response from the Nonstop compute community is positive and very supporting. As for responses from the mainframe community, up until now, there’s been nothing. The post was made in response to IBM marketing talking up the high level of nines so for them to stay as quiet as they have is telling. Safe and secure? Perhaps this isn’t the storyline you expected to hear from me but in the telling, I believe the Nonstop Compute community should be very proud of what their latest Nonstop Compute converged system provides and in no way, should they fear any confrontation from any of their IBM peers.

There’s nothing new under the sun for someone who has been in IT for as long as I have, when it comes to availability – the very essence of why HPE continues to invest in Nonstop Compute as it delivers extreme levels of availability – it still means something very important to all CIOs. A night of sleep filled, uninterrupted, rest. With that, the Nonstop community can safely say, Nonstop has secured such rest, day after day, for the four plus decades of its existence. That in itself should be enough to awaken any CIO to the true valuation HPE now places on all things Nonstop.