TEDOG. That’s how I think of the Disruptive Technology track at Rice University’s annual Oil & Gas HPC Conference. That being the case, the idea I chose to spread at this year’s event had to do with containers – open platforms that allow developers and sysadmins to build, ship and execute distributed applications.
Because early adopters like The Centre for Genomic Regulation (CRG) claim “easier replication, faster deployment and lower configuration and operating costs”, my objective was to encourage Conference participants into a hydrocarbon-centric dialog on containers. I expected this dialog would be timely as Univa knows of numerous organizations that are already investigating containers via Docker in proof-of-concept and/or pilot initiatives. Our experience has shown, however, that successfully transitioning these initiatives into production deployments can be fraught with obstacles.
To stimulate the conversation, I seized my TEDOG moment by delivering the three-minute presentation captured here and poster here . After providing a little background on containers, I shared our preliminary conclusions. One of the key conclusions is that Univa Grid Engine – Container Edition allows Docker containers to be managed as workloads, alongside other types of workloads, in an HPC cluster. Feedback we are receiving from early adopters of this solution indicates that it is a key enabler in successfully transitioning to production deployments from investigatory initiatives.
At the time of the presentation in early March, I was fairly conservative in stating preliminary conclusions regarding support for MPI applications as well as GPUs. I am delighted to report that I can be considerably less so today:
Finally, reference was also made to cloud-native applications in the three-minute presentation as well as the poster. Briefly, Univa is working with numerous other organizations on a complete re-think of applications in the cloud via the Cloud Native Computing Foundation (CNCF) – a re-think based around containerization, dynamic scheduling and an orientation around micro services. For more on this, including the ability to provision a Kubernetes container cluster 5 times as fast as other solutions, please review our Navops offering.
Despite the current state of economic downturn in the oil and gas industry the 2016 Rice University Oil & Gas HPC Conference surpassed all previous attendance records. While this industry remains embroiled in such a state, proactive organizations are looking to disruptive technologies such as containers. If containers factor into your plans in any way, we’d certainly appreciate hearing from you.
Shameless plug: In answering the 5 Ws and one H, the aim of our Docker 101 webinar is to provide a technical overview and demonstration of Docker and to frame its use within the context of High Performance Computing and Big Data Analytics. Register here.