2021. What an interesting year. With the world turned upside down by a pandemic that seemingly had its sights set on...
The HPE Corner
HPE Discover Madrid revealed much…
From Synergy to Superdome Flex, the support of commodity hardware and industry standards means NonStop has even more of the HPE landscape to address!
As last month’s update on all things HPE came to a close, reference was made to promotional posts made by Kirk Bresniker, HPE Labs Chief Architect, in support of his presentation on The Machine that he would be giving at this year’s HPE Discover event. Coming as soon as it did following a highly successful NonStop Technical Boot Camp, there was sure to be a lot more than simply passing interest in what HP Labs have been up to in the intervening years since then HPE CTO, Martin Fink, took the wraps off The Machine. For the past couple of years you could include the NonStop community among those who were giving The Machine only passing interest but as the sessions began piling up at HPE Discover, it became very clear that the mantle of The Machine has been thrown across almost every development program.
The invitation to be a guest of HPE Discover came late for many of HPE’s designated “influencers” but all the same, when the final count was tallied, most of those associated with social media turned up for this biggest of all big-tent events. Bloggers are always in need of fresh content and nothing slates that thirst better than HPE Discover. From the very first breakfast session to the final photo op, HPE did a really good job of ensuring a variety of executives and managers made it to the influencers’ conference room. While this group had the opportunity to do pretty much anything they wanted, the program for them was well thought-out so few bloggers elected to miss any of the sessions. And yes, Kirk was walking the floor of the Transformation Zone happily answering questions about The Machine and pointing to the new logo for The Machine – yes, the three green bars about which you will be hearing a lot more in the comping months.
The Machine’s impact could be seen in the just announced Superdome Flex servers as well as in Synergy frames. It was just a short time ago that HPE introduced Superdome Flex but the more time spent reviewing the specs – combining technology from Superdome X with SGI’s UV300 (now called MC990X) – reveals both the scalability on offer and the obvious real world introduction to memory-driven computing. As one blogger posted to the HPE web site, “While the memory in a Superdome Flex system is not physically separated out onto the fabric as in The Machine prototype, with small changes in the operating system, it is possible to make the application believe that it is running on The Machine (within the limits posed by 32 sockets and 48 TB memory);” 32 sockets and 48 TB of memory being the upper limit of a Superdome Flex configuration today. Furthermore, “This development provides our customers and partners with yet another opportunity to realize the benefits of Memory-Driven Computing right now, and is another example of how HPE is leading the industry with cutting-edge in-memory computing solutions.”
OK, so for the NonStop community what is strikingly clear is that this isn’t a refresh for the HPE-Unix community (it runs Linux) but rather, a tangible insight into where HPE is headed in terms of servers within its product portfolio. It was hard to miss references to such well-known hypervisors (KVM and yes, VMware) as well to the optimization being undertaken to ensure a future leadership role in a fully virtualized world. Running mixes of virtualized workloads seems to be its sweet spot and among the early real world use cases happens to be SAP Hana. Yes, analytics and the need for better streaming analytic platform support is very much in the sights of the developers even as HPE itself is looking to its partners to support such streaming analytic platforms among which, vendors known to the NonStop community (including Striim) will likely benefit. Among the very earliest users of The Machine and its prototype has been the German Center for Neurodegenerative Diseases (DZNE) and the University of Bonn who are looking “to speed up their research into brain diseases like Alzheimer’s.”
At first, the researchers in Germany “got their hands on two HPE machines this year: the Superdome X and the Memory Fabric Testbed” but now they have been upgraded to Superdome Flex. Performance gains of 100 times previous performance levels were achieved. Among the blogging community there was some recognition of the irony that comes with the HPE approach to memory-driven computing being applied to diseases associated with human memory. Another big winner when it comes to exploiting memory-driven computing was observed when processing financial models, including Monte Carlo simulations, where such models ran 10,000 times faster. Again, important when you consider the path from the intelligent edge into the cloud will necessitate passage through analytics platforms in real time!
When many vendors are electing to focus on the “D” of R&D it is encouraging to see just how big an investment HPE Labs continue to make into “R.” For the NonStop community the implications are clear – the move to commodity hardware and the evolution of NonStop to software with the L Series capable of running practically anywhere HPE elects to support it, NonStop is joining a growing list of virtualized workloads. Simply provision a virtual machine and NonStop can be loaded – yes, Superdome Flex is x86 and RoCE equipped so the basics are all there if you have been wondering whether NonStop could run on Superdome Flex.
Now, I am not saying that this is likely in the near term and to date, the Superdome Flex is more in tune with the needs of applications that need to scale up (and enjoy the performance improvements that come from doing so) whereas NonStop is more attuned to scale out but then, you never can tell. But if you have any interest to do so and can build a strong use-case in support of such a configuration then just let the HPE NonStop team know about it as there are only a few knobs that need turning in order for this to be a reality. There may be a little head-scratching among NonStop traditionalists but then again, there is so much inherent power in this architecture it will likely generate some interesting exchanges as I am sure, in time, there will be those applications that elect to scale out on even the most basic of memory-driven computers.
And this is the biggest message of all coming from HPE – NonStop is core and yes, NonStop is high value. These are two very critical elements in ensuring R&D continues in support of NonStop and when you think about the work being done in support of NS SQL/MX and think too of NonStop’s position (and value proposition) as an intermediary linking the intelligent edge to the cloud, then you can see why there continues to be as much excitement surrounding NonStop as there is today. While individual components within The Machine, as it is currently being deployed, reach into the high 9s of availability, HPE continues to pull back from any suggestion of fault tolerance and yet, when it comes to the real world of the Edge and the Center, this is exactly what end users will be looking to deploy and for that, across HPE, NonStop has no peers!