Between ISC and HP-CAST (held Friday and Saturday this year), I had several speaking opportunities, wherein I mostly talked about our release of Moab 8.0, and the many new features and improvements it provides to our customers. However, participating as one of the panelists on Disruptive Technologies, I had the opportunity to discuss where I see the industry going.
As the only panelist representing a software company, I thought it was appropriate to provide a slightly different perspective than the quantum computers, GPU-enabled ARM systems and other such things being discussed by the other panelists. Contrary to what is often talked about in a tech-heavy, pure science conference, I wanted to talk about us… about our community… the HPC community.
Disruption is always around us. The rate of technological change is breathtaking, with “newer,” “better” and “faster” always being the watchwords of the day. But, aside from that, there is another, more subtle disruption that is occurring: other areas of computing are adopting tenants of HPC and HPC is being infiltrated by ideas and practices from other areas of computing.
This is not a bad thing.
In fact, the cross-pollination of ideas is, in my opinion, a necessary step in allowing the different computing technologies to continue to grow and evolve. Coming from sometimes very different points of view, HPC, Cloud Computing, Technical Computing and Big Data each play an important role in general computing, which we can define as computing specifically designed and used to find answers and insights to questions. Granted, that’s a pretty wide net to cast, but it doesn’t, for example, include using one’s workstation to pwn some noob in Call of Duty.
The question really then becomes whether or not we of the HPC community will accept these new influences. Personally, I don’t think we really have a choice, but we’ve been known to have rather heated arguments over what and who constitutes HPC, as if the designation represents membership in some hoity-toity country club. We draw some arbitrary line and say if your computing power is above this line, you are doing High Performance Computing. If you aren’t “high” enough, I don’t know what we call you… Medium-High Performance Computing (MHPC) perhaps?
Oddly enough, this line moves over time, and what once was considered HPC no longer is, and the CRAY-1 supercomputer becomes the visitor’s waiting bench. Why not? It was built with a seat surrounding it.
However, times have changed, and just has HPC’s sphere of influence has increased, so has our need to be inclusive as a community. We are moving towards a day when computing tasks of different types will be strung together in operational workflows. Consider the following sequence:
- An engineer logs into a portal to request a virtual workstation provided by the cloud.
- Using the workstation, she does the rough design for a new product, but doesn’t have all the details for the target demographic. So she starts an automated workflow.
- Using parameters provided by the engineer, a Hadoop job is started to determine biometric data for the target market.
- Once computed, the biometric data determines the values for the parameter sweep to be done by a series of HPC jobs.
- The results are collected and sent to the visualizer.
- Once this is done, the engineer collaborates with her other colleagues, all who are remotely accessing the visualizer.
This fictitious, yet plausible, scenario is one I believe is indicative of what we will in there future. It’s a workflow designed to use the best tools for each step to get to the answers and solutions as fast as possible. It’s what Adaptive Computing calls Big Workflow.
And, it was this future I talked about in the panel discussion. Granted, at times it felt a bit like an after school special—a television show from my childhood that often dispensed pithy “everything will be all right” homilies on what were actually really important issues of the day for teens such as drug and alcohol abuse and teen suicide. I even had to mention it during one of my “we all need to play nicely together” moments, which elicited a laugh and snicker from the audience.
But, I do believe that we need to “play nicely together” if we are to succeed as a community. Otherwise, we’ll be passed by as the newer computational technologies mature.
So, my reason for concern is also my reason for my excitement, and the excitement far outweighs the concern. There are exciting days, months and years ahead of us in the world of HPC. I predict they are only going to get better, as we move forward.
Come along for the journey!