HPC Unlocks Innovation
CEO of Intersect360 Research, Addison Snell, discusses
HPC’s ever-changing landscape, its hottest tech trends,
and what’s exciting him most about the market.
5 June 2016 www.HPCSource.com
A: Technology also goes in the other direction, from general
enterprise computing into high performance
computing, most noticeably, in areas having
to do with virtualization or cloud.
Containers have been around in
enterprise computing for a while.
But now that Docker can expand
Q: How has high performance computing looked at
the sense of a container to where
it represents an entire virtualized
cluster with its own application
environment, you’re seeing that
migrating over into high perfor-
mance computing and cloud has,
of course, been a major trend across
enterprise. And the high performance
computing user is figuring out when it
makes sense to not just have a private
cloud but access to a public cloud virtualized
environment. We see a third of the industry now incorporating aspects
of public cloud into their high performance computing environments.
massive amounts of data or different forms of cloud
computing for years?
A: When things like cloud or big data get hot and begin taking
over enterprise there’s always a bit of a notion from some of the
supercomputing centers that this isn’t new—we’ve been doing
utility computing since the 1960s; we’ve got mountains of data for
our scientific endeavors. But that’s true—they’ve been around for a
long time, but you don’t want to get too complacent in the idea that
just because it’s now being done in the industry in a new way that it
isn’t also new in a different way.
We have new types of data, more types of data that are arriving,
even if they’re at a lower level of scale than at a supercomputing
center. And you often see technologies that come out of these
enterprise deployments in cloud, in big data that then start
migrating back the other way and become incorporated in the
large, supercomputing centers so that there’s a continuous feedback
loop of new ways of thinking and best-of-breed technologies. Is
cloud really new? Well, utility computing isn’t new but the idea of
how you virtualize that environment and access it in a new way—
that is a new way of thinking about a problem.
Q: What are some of the hottest technology trends
in HPC today?
A: There are always important new trends on the forefront
of high performance computing. One today is looking at
the effect of hyperscale markets into high performance
computing. When we talk about hyperscale, we’re
looking at the ultra-scale Internet kinds of companies—
the Googles, Facebooks, Microsofts, Amazon, Alibaba
in China—and how they’re taking their webscale
application infrastructures, which is not quite the same
as high performance computing science or engineering
application, and achieving that level of scale.
Now we’re seeing technologies that go not only from
high performance computing into hyperscale, but how
those levels of innovation in hyperscale work their way back
to high performance computing. Most important right now is
the areas of machine learning, or cognitive computing, in which
you have artificial intelligence, or applications that assign all of the
data you have at your disposal, to get the computer to learn patterns
in that data so the computer can help you determine that next level
Q: Why is machine learning such a big deal, if it’s been
around for some time?
A: The idea of machine learning, deep learning or artificial
intelligence isn’t new; it’s been around for a long time. But the levels
of scale at which it’s being deployed at some of the larger hyperscale
institutions currently, like Google or Microsoft, is really being taken
to a new level. You’re seeing advancements in machine learning and
cognitive computing in ways that weren’t previously employed. For
example, Apple’s Siri or Amazon’s Alexa, which are able to respond to
you in a natural language, or an image recognition web search where
you can ask a question, ‘What am I looking at?’ That’s something
that a search engine hasn’t been good at before unless the picture is
specifically tagged that it’s a bird sitting on a branch; the computer
didn’t know it was a bird sitting on a branch—now we are getting
where it does and that’s important for the new world of data we live in.
Q: What examples of enterprise technologies
have been adapted in new or different ways, in high
performance computing or hyperscale? Does it ever
go the other way?
to remain continually
focused on the industry’s
future and that next, new
technology to empower
on an ongoing