Virtualization Technology News and Information
Article
RSS
Nimbix 2019 Predictions: 2019 will be the year Linux containers actually work for HPC and scale-out deep learning

Industry executives and experts share their predictions for 2019.  Read them in this 11th annual VMblog.com series exclusive.

Contributed by Leo Reiter, Chief Technology Officer, Nimbix

2019 will be the year Linux containers actually work for HPC and scale-out deep learning

Before I get into my 2019 prediction, let me first gloat about my 2018 one: it happened.  I boldly predicted that 2018 would be the year the cloud finally democratizes FPGAs.  Maybe I had some insider knowledge or maybe I didn't, but it was a big year for FPGAs in the cloud.  Not only that, but we made it almost "too good to be true" by partnering with Xilinx to offer up to 100 hours of free access to their Alveo cards and development tools in the Nimbix Cloud.  If you're interested in FPGA development on the latest and greatest technology, it takes seconds to get started with our Alveo trial.

As you probably know, Nimbix has been in the democratizing accelerated compute business for almost a decade.  We were first to market with on-demand GPU cloud computing, and we also invented what's now known as "Reconfigurable Cloud Computing," which in large part refers to programming accelerators (like FPGAs) as part of workflows.  All we did in 2018 was make it dead simple and took cost out of the equation for those looking to get their hands dirty with some of the most exciting technology ever seen in the public cloud.

So now that I've proven my clairvoyance, I'll stop gloating and talk about what I think will happen in 2019.  Not surprisingly, Nimbix is leading the charge here as well, but it began years ago.  Before there was Docker or Kubernetes, Nimbix was delivering containerized workflows in the public cloud and innovating on things like acceleration and tightly coupled massively parallel architectures in this space.  Last year we decided that Kubernetes was gaining enough traction in the cloud and the enterprise that it was time to bring HPC to it.  HPC is not a four lettered word - it's actually 3 letters, and stands for High Performance Computing.  Over the last few years, HPC has played a pivotal role in digital transformation, and Nimbix has been there democratizing things like advanced simulation and scale-out (or distributed) deep learning.  We've been doing it in Linux containers, on bare-metal, from (almost) the very beginning.  We developed a cloud operating system and application delivery platform called JARVICETM, and we took it to market as a service.  In the last two years alone, JARVICE in the Nimbix Cloud has powered over a million containerized supercomputing workflows - in many cases, containers spanning dozens of high-density compute nodes (or more) - we still count that as "1 workflow" though!  So yes, over a million answers to some very complex questions. 

Our customers have enjoyed bare-metal performance, seamless access to the latest computational accelerators, and turnkey simplicity without even having to think about things like container orchestration or runtime engines.  Thanks to the SaaS delivery model, our users don't even need to learn how to spell YAML, much less write it, to get their work done.  JARVICE 3, our first true hybrid cloud platform release, works on virtually any Kubernetes cluster, bare-metal or virtualized, and deploys in minutes. What's more, our proven service catalog of over 1000 commercial and open source workflows behaves like an "app store," automatically synchronizing to on-premises and remote cloud clusters, so users don't need to worry about application deployment, period.  The same scale-out point-and-click simulations, neural network training, and other compute-intensive workflows are now available on any cluster, and virtually any cloud.  Of course, for best performance, it's hard to beat the Nimbix Cloud, due to its vertically integrated supercomputing powers, but we understand that latency, regulations, and other variables sometimes dictate running compute elsewhere.  Thanks to JARVICETM 3 and Kubernetes, HPC and scale-out deep learning is a first class citizen again.  Customers can adopt a converged IT strategy and settle on a single infrastructure management platform, and JARVICE does the rest.

Which brings me to my 2019 prediction... Now that it's so seamless to run containerized HPC anywhere, I'm going to go ahead and predict that we'll see more and more containerized infrastructure delivering these workflows, even if outside the Nimbix Cloud.  And because our customers demand technology that "just works," we'll expect the market at large will start demanding that as well.  If you're curious or have a tough problem to solve, contact us, and we'll be glad to show you how.

##

About the Author

 

Leo Reiter, Chief Technology Officer, Nimbix

Leo is a virtualization and cloud computing pioneer with over 20 years of experience in software development and technology strategy. Prior to Nimbix, Leo was co-founder and CTO of Virtual Bridges.
Published Monday, January 21, 2019 8:02 AM by David Marshall
Comments
There are no comments for this post.
To post a comment, you must be a registered user. Registration is free and easy! Sign up now!
top25
Calendar
<January 2019>
SuMoTuWeThFrSa
303112345
6789101112
13141516171819
20212223242526
272829303112
3456789