So… I got exposed to this when we started more and more HPC work at EMC. Isilon is a fit for some workloads, Lustre/VNX is popular for others, and then we did some amazing stuff with Los Alamos Labs regarding developing use of Flash as a “burst buffer appliance” for Fast Big Data use cases (were Petabytes need to be checkpointed in seconds/minutes). You can see Los Almos National Labs (and Wade and I) talk about it at Chad’s World Live II – “the comeback tour” at EMC World in Las Vegas. It’s MIND BLOWING. Think of “million compute core” infrastructure – moving to “billion compute core” scale and Exaflop-scale… and you’re getting the idea of the scale.
Fast forward to 44:30 into the episode below:
As part of the “Big Ideas” series (there were earlier ones on Hadoop and Big Data), Patricia Florissi – Americas CTO and general awesome person at EMC did a couple of videos around the topic of High Performance Computing.
We use these videos we use internally at EMC to bring people up to speed and understanding these “Big ideas”. While it’s what we use internally here at EMC, virtualgeek readers know my view is that we should share openly and freely.
I think these two videos do an AWESOME job of making complex ideas, and BIG ideas come to life – and learning stuff is one of the things I take a great deal of pleasure out of… after all the EMC Presales Manifesto has as one of it’s principles that we are technologists, and technologists are intrinsically curious, and always independently learning….
Here’s the first HPC Big Ideas video (click): “Simplifying HPC Architectures”…
.. And here’s the second HPC Big Ideas video (click): “Demystifying Fast Data” (which looks at the Los Almos challenge which is common to many large HPC environments which are massive PB-scale checkpoints, ergo the ABBA idea)
If you dig these, check out all the “Big Ideas” videos here. Enjoy, and feedback welcome!