Tuesday, December 15, 2015

Picking the line …

Following the interviews and discussions that took place during the 2015 NonStop Technical Boot Camp it was time to follow-up with those working more closely on the products featured in keynote presentations. And the storyline that develops here takes it cue from those discussions …

With the arrival of winter entering our garage is tinged with a degree of sadness with wires snaking around cars as they stand connected to battery charges. Not for the icy roads of Colorado are these rear-wheel drive sports cars and while we continue to transition to more suitable vehicles for our location, I cannot help but anticipate sunny days ahead even as I save away a few pennies for new tires on our track-focused sports car.

In any highly competitive car race, where there’s barely any separation between competing drivers, there’s little chance any driver will be able to hold the perfect line around the course. There’s simply too many competing for too few feet of track and every turn a circuit throws up at the competitors sees drivers prepared to try different lines to gain an advantage. As I was once advised, “observe and learn the racers line around the track as when it comes to the event, you will likely never be in a position to follow it.”

Point being, in a close field, every driver has to have something in reserve to counter what those around him do and while everyone is close to the edge (of disaster) the good drivers, and those who are perennial winners, just know when to feed in the corrections needed to avoid disaster while gaining that elusive winner’s edge. In IT, we call this transformation in the sense that we have to walk the line between what it is we need to support – those production applications that cannot afford to be disrupted – and where we need to be. Furthermore, for each step we take that moves us closer to the transformation goal we pursue, we need to ensure what we already have in place inches forward right along with the steps we take.

This past week has been tumultuous to say the least. Put it down to the proximity to the holiday season, or simply the end of a quarter or even end of year, but the type of inquiries being made and the number of questions being asked has been as high as I can recall in all my time as a blogger. Whether it is a private email or a more public exchange on one of the more popular online forums, there’s no questioning the NonStop community’s enthusiasm to participate in the debate over the future of NonStop and the direction it is clearly taking. Following a number of Hewlett Packard Enterprise (HPE) big tent events worldwide, including the NonStop Technical Boot Camp, the HPE spotlight has intensified its beam directed at NonStop.

At Boot Camp we heard from Martin Fink, EVP and CTO, Hewlett Packard Enterprise, Randy Meyer, VP and GM, Mission Critical Systems at Hewlett Packard Enterprise, and Andy Bergholz, Director of Engineering, Hewlett Packard Enterprise, where the outcome was certainly one that encouraged all who participated in the event. Not for a very long time had there been such a resounding message of support of NonStop given to the NonStop community and the fallout from the keynote presentations continues to reverberate through the NonStop community even now. For many of us it was a wake-up call to truly reconsider the future of NonStop systems as HPE has clearly made a sizable investment in its future even as it turns up the heat on the value proposition of NonStop systems – there’s little confusion or ambiguity about the intentions of HPE concerning NonStop. “NonStop X? It’s the best software platform on the planet!”

When Martin Fink took to the stage at Boot Camp, he walked the audience through the moves that were required to execute on a strategy and then projected a vision for NonStop unlike any we had heard previously. It was more than just the separation of NonStop from a dependency on hardware. It was more too than just the potential for NonStop to have an impact on developments taking place in areas such as clouds, virtualization and even the open source community. As Fink noted, “Adoption of open source is not a technology choice but a business choice,” and a reflection on changing attitudes within IT.

“Take NonStop to open source? Take a NonStop feature to open source,” Fink suggested at one point, even as he posed the question of whether HPE could, “Enhance existing projects with NonStop?” In other words, HPE could pick from many lines to take when it comes to NonStop, but even so, with other powerful interests all around HPE seeking the same track space as HPE, it was going to take a lot of concentration on the part of HPE engineering to steer NonStop clear of the competition.

However, when Fink touched on virtualization he touched a nerve with everyone in the audience. “Running on a virtual (environment) on a Linux,” Fink asked before adding, for those interested in the topic, “As an important proof point, we can absolutely get there.” Furthermore, looking at it a little differently, “Wouldn’t it be cool to bring the NonStop value proposition to Linux and bring to market (more) powerful hybrids – a powerful combination.” If there were to be a future for NonStop in clouds, for instance, then there has to be a future for NonStop atop a virtualized world.

Fink closed his keynote suggesting too that the NonStop community needed to consider the possibilities of a future where NonStop could be supplied as a service as well as NS SQL. Tantalizing, for sure, given the task of engineering NonStop to run on a variety of virtual machines atop of popular hypervisors – something that Fink alluded to as already happening deep within the NonStop engineering labs (as evidence of an important proof point being reached) – and yet surprising news all the same.

In the weeks that have followed Boot Camp I have been able to catch up with Andy Bergholz and if not immediately apparent to those attending Boot Camp, the source of much of Fink’s renewed enthusiasm for NonStop came about following discussions with Bergholz in the days leading up to Boot Camp. There have been numerous projects announced by HP in the years prior to the formation of HPE. Project Odyssey, Converged Infrastructure, and more recently, Composability and even the most recent announcement pertaining to Synergy and Docker. But the work on NonStop to firstly remove dependence on the hardware even as NonStop embraced the Intel x86 architecture has its roots in work done long before these projects received the attention in the press that they did.

“NonStop is not just about being independent of the hardware,” said Bergholz. “The goal for NonStop is to make it independent of the infrastructure; as we see it NonStop is above the infrastructure and our goal truly is to include within NonStop an infrastructure independence such that you can layer NonStop atop your choice of infrastructure.” As for markets, Bergholz acknowledged that “the ultimate goal of NonStop in a converged infrastructure market (so that) you can continue to buy NonStop systems and I want to be able to sell as wide as possible in terms of verticals. But there is another customer base that want to run clouds and NonStop needs to be able to run seamlessly across their chosen (cloud) infrastructure.”

And yes, they have NonStop running on top of at least one hypervisor without the need to add anything to the hypervisor and that, Bergholz added, “We have no plans to require changes to the virtual machine a customer selects even as we want to be able to run atop as many different virtual machines as may be out there.” Part of the work being done by NonStop development is to come up with a reference architecture whereby obvious configuration requirements will be provided – don’t configure two NonStop processors on the same CPU, for instance. “Our goal is for customers to fully understand the (likely fallout) from the configuration decisions they take.”

For some time now, the industry has accepted the definition by IDC, “the premier global provider of market research” of Availability Level 4 (AL4) and a reference to a level of availability only NonStop along with select implementation of IBM’s mainframe Parallel Sysplex can provide. Imagine a future virtualized environment, cloud or otherwise, where checking the box for AL4 brings with it the provisioning of a NonStop as a Service capability. Likewise, check the box for mixed workload SQL whereby tables can be simultaneously access online by transactions even as statistics are performed and the maintenance of tables is pursued, and NS SQL as a Service will also be provisioned.

For anyone who has been on a circuit competing with likeminded car enthusiasts, finding and maintaining the racers line is never an easy task. The challenge in doing so often calls for minute adjustments that take the car away from a preferred line but in so doing, the skilled driver can gain the upper hand. The course that has been followed of late by NonStop development hasn’t always been clear to the NonStop community but having been given the scoop on strategy, vision and now goals for NonStop, the picture is becoming a lot clearer. Avoiding potentially disastrous race-ending “incidents” is a priority for all racers and this is clearly something HPE engineering has been every bit as sensitive to as the most seasoned racer. And having said this, NonStop has picked the right line and it’s only clear track ahead!

1 comment:

Dean E Malone said...

Great article Richard. It will be interesting to see how they leverage the distributed shared-memory capabilities of IB to implement a fault-tolerant Universal Memory framework.