HPE is committed to becoming a significant player in edge to core; will we see NonStop contributing to both edge and core?
It was just
about two years ago that Margo and I spent a week in Munich. It was kind of a
rushed trip but we wanted to catch up with former business colleagues from the
time when we both worked for the former Nixdorf Computers. That was the time
when Nixdorf made a full range of products from key-to-disk data entry systems
to retail and banking terminals, including ATMs, a PABX system and yes, even a
plug-compatible mainframe. This I covered in the post of March 31, 2019 Living on the edge, a post that was widely read at the
time.
Not sure if it
was the reference to Munich, to wintery conditions, to Nixdorf and IBM or
whether it was the reference to edge computing but it included an important
reference to quotes of the day by HPE CEO, Antonio Neri that continue to
resonate with Margo and me to this day. “Edge computing means that ‘millions of
clouds will all be connected,’ said Neri. ‘Every edge to every cloud. That's
the future and why I'm excited about the future of the company.’” Can’t make it
much clearer than that or hide the fact that indeed Neri was excited. Without
making too fine a point about this, Neri continues to be just as excited today
about the edge to cloud message as he was back in 2019.
In the few
discussions that I have had with NonStop vendors of late it has become quite
clear that talk of moving processes and data to the cloud is just that; talk
and little else. Not entirely unexpected, as cloud service providers really
struggle to price effectively solutions that need to run 24 x 7 and that need
to have the database online the whole time. Simply equating running on external
clouds as being little more than running a solution on someone else’s hardware
doesn’t help strengthen an argument in favor of clouds and yet, it continues to
resonate well within many enterprises.
I cannot hide my
observation that even with private clouds that are beginning to take shape
within the data centers of some enterprises, running mission critical solutions
that today run on NonStop within a private cloud isn’t a priority. And why
should it be? For these enterprises, they have the skills needed to run NonStop
– can they be assured that the same level of expertise will be directed at the
cloud private, or public?
No, the more
relevant question is not should I cloud or should I stay but rather, what can
the cloud teach me? If you follow Neri’s headlines of date it’s not so much
about the cloud but the cloud experience. Enterprises would welcome as much
flexibility and dare I say simplicity that they associate with running from out
of a cloud being applied to all their traditional systems, including NonStop.
This will remain
under debate for some time even as I am conscious of the fact that few
migrations to even the most robust of private clouds will happen for NonStop in
2021 or even 2022. More focus will be given to whether the time is right to
upgrade to the new NonStop NS8 X4 (or NonStop NS4 X4) than what might be given
to moving to a virtualized cloud environment. For good reason, the average
enterprise invested in NonStop is risk averse – take down the most critical of
mission critical applications and face the consequences? Not likely and who can
say; will there be any cost benefits even if it all goes well?
What may not be
debated for too much longer is whether NonStop should move to the edge. There
is a certain elegance about placing processing power as close to user
interactions as possible. Lots of processing power to where there is a
concentration of user interactions – robots, warehouses, ticket offices and the
like. The elegance comes when you consider that the database is a candidate for
the core, cloud or traditional, and the processing a candidate for the edge.
Today NonStop is
blessed with multiple solutions that deliver distribution through replication
solutions so maintaining currency edge to core is not that big a deal anymore,
bandwidth permitting. 5G deployed, already? Vendors like Striim and even NTI
come to mind as they rely on Change Data Capture (CDC) methodologies which lend
themselves to better integrate data with external databases and processes. It’s
very easy to see why they welcome the opportunity to distribute data where the
topology revolves around this every edge to every cloud that Neri referenced.
What may tilt
the argument more strongly in favor of NonStop at the edge will be AI – giving
the local processors insight into user interactions in terms of changes in
behavior would benefit from NonStop systems ability to keep processing 24 x 7
even as system managers would maintain an ability to update local databases
with no downtime. Today’s NonStop systems can even grow to accommodate greater
traffic volumes without downtime and when it comes to pushing security as far
away from the core as possible, wouldn’t you really have more trust in a
NonStop out on the edge more than anything else?
Alternatively,
some enterprises might consider operating a lot of deployed NonStop systems as
being beyond their skill sets. After all, there would be a lot of consoles to
oversee and perhaps it might not be practical to staff every console required
of NonStop out there, on the edge. However, is this really the case these days?
Have we not progressed our monitoring and management capabilities to centralize
such tasks? Talking to service providers such as TCM Solutions, it would appear
that this is already being undertaken to some extent as such service providers
can already oversee multiple NonStop systems that are geographically dispersed.
Let’s get
serious now just for a moment. NonStop never has had a small configuration that
was in any way meaningful. There was the original “lower cost” NonStop CLX that
met with some success but not over the long term. Ever since development demoed
the NS2 “desktop” NonStop systems a few years back – a system that was
virtualized and ran VMware (or was it the Openstack virtual machine, KVM?) but
again, I haven’t read about too many wins of late for this desktop system. If
we consider the edge almost by definition will be a couple of ProLiant servers
supporting virtual machines then maybe, just maybe, we have a place to start.
HPE launches Edgeline EL8000 Converged Edge
System
What about the
Edgeline servers from HPE – the 4000 and the 8000 models? Haven’t there been
demos of them running NonStop on virtual machines? Yes, I have seen examples
myself where NonStop was present but within HPE little traction has developed.
Not that I have come across. I don’t want to position these Edgeline offerings
as some kind of panacea but rather to suggest that in a virtual world, where an
appropriate virtual machines can be found, therein lies opportunity for
NonStop.
So, can we feel
good about a future that has NonStop running on the edge or is this still a
case of it looking good on the whiteboard and in PowerPoint but seriously,
problematic at best? In a world where HPE CEO Neri has also committed HPE to offering
every product in its portfolio on the basis of as-a-service, could this prove
to be a way out for NonStop on the edge?
Following its launch,
possible sighting of a rare NonStop NS2?
Lots of
questions to be sure but addressing the potential for NonStop at the edge isn’t
something we leave to HPE and the NonStop team. We really cannot expect them to
push deeper into the solution space. If
as a community we want to feel good out on the edge then it really is up to us
to identify opportunities and then build-out solutions that address the
requirements we come across. Any interaction between an end user where value is
returned to that user, such as is the case with looking up a QR code to see a
menu, warrants a NonStop solution. Point is, do we really want to flood our
networks, 5G or otherwise, with what we have all heard described as digital
exhaust?
Debates will continue over whether it is feasible or indeed practical to move NonStop out of the core and to the edge. It’s too hard to manage and yes, it’s difficult to physically size appropriately. However, processing will continue to move out of the data center to become positioned at the edge. Is it our intent to let such a generational pivot be ignored and to see NonStop miss out on one very large opportunity? There will always be a need for operational availability NonStop alone provides and for that, we should be attuned to just how good it might be to process out on the edge.
Comments