The HPE Corner

There are a lot of challenges facing HPE in 2017. This shouldn’t come as a surprise to anyone familiar with all that is transpiring within HPE, but there’s a lot to be excited about as well. First, the restructuring going on within HPE as the Services group was sold off to Computer Sciences Corporation (CSC) and then only a matter of a few months later, non-core software was sold off to MicroFocus. There may be those within the NonStop community puzzled by these sales but it’s time to get over it and move on.

Wondering what security was viewed as a non-core asset may be a good topic for in-depth discussions over beers, but in all seriousness, it’s not like it was abandoned. All the non-core software assets have simply found a new home and one where HPE will still hold a 50+% interest. So, yes, all should be cool on that front.

Excitement about HPE really comes when you pull apart the product mix within the Mission Critical Systems group headed by our own Randy Meyer. Stepped as he is in all things NonStop and happy with double digit growth of NonStop over the past two years it’s no surprise that his focus lies elsewhere for now. However, the purchase of SGI and its contribution to High Performance Computing cannot be ignored – this is perhaps the “sleeper deal” of the decade and will help add a lot of new revenues that will bolster HPE’s bottom line for servers.
But the challenges persist. No surprises here for the NonStop community. Recently I heard about a NonStop user who was advocating how they could move to a commercially, off-the-shelf, server platform and with their knowledge of NonStop, build their own “non-stop!” How ridiculous! Yes, you can do that but all you end up with is something that is almost as good as a NonStop – but no, not a NonStop. And why? When you look back at the NonStop Availability (NSA) initiative of the late 1990s, headed by Dr. Tim Chou, by the way, you will appreciate that contributions to failures come from a combination of deterministic, nondeterministic, operational, environment / infrastructure and yes, end user.

Ignoring the end-user aspect as our solutions shouldn’t crash because someone hit the wrong key and yes, when it comes to operational there’s always the recognized need to keep two Dobermans on station between the console and the operator – primary and backup, I suggest (and remember that scene from Moonraker?) – but accommodating both deterministic and nondeterministic failures well, that’s particularly tricky. There’s a reason why NonStop Systems were never “top vented” – ask anyone who was involved in the early design of NonStop systems.

The point is, you can almost build a NonStop system using any messaging system but it’s a case of almost. What’s missing is the decades of experience accommodating everything that can go wrong. And that too is now the challenge for HPE with NonStop. Dismantling the argument, “Good enough is well, good enough!”

For NonStop to flourish in the 2010s and beyond, HPE needs to give more serious consideration to the true cost of downtime. No, you cannot build an equivalent to NonStop and no, you cannot discount the value of true fault tolerance. Even today. Just ask anyone who has had plans disrupted because of technical glitches at airline’s computers – it’s just not acceptable. And at the very heart of the challenge for HPE is to aggressively dismantle the cost of ownership – it’s truly a myth that NonStop is more expensive or that it needs special skills.

Somewhere out there a NonStop user has successfully deployed MineCraft or the equivalent on NonStop – it’s just not a secret worth keeping that you need specialist to support NonStop. The challenge for the NonStop community lies with our ability to stand up and defend and indeed promote NonStop as being THE mission critical system. No misgivings, mind you, and no holding back, and it really is an easy message to communicate these days – and you have a highly vocal and motivated community at your beck and call to support you if you need help.

Yes, there are challenges for HPE but there is also the realization that NonStop is entering a new “golden era!” Want to buy the whole physical system from HPE? Check! Want to buy your own x86 servers from a white-label manufacturer in the Far East and run vNonStop? Check! Want to run your solution from out of a cloud populated by NonStop? Check! When was the choice of NonStop options this rich? Yes, HPE has challenges when it comes to promoting NonStop but when it gets past them, it has a lot to be excited about!