Univa Grid Engine Takes On Docker Orchestration, Scheduling
Workload management and optimization solutions developer Univa is entering the Docker orchestration frenzy with today's release of the Univa Grid Engine Container Edition product.
In September 2014, Univa told EnterpriseTech of plans to add Docker support, predicting it would arrive March 2015. But Univa invested more time speaking to customers, researching the market, developing a solution, and talking to partners like Docker and Google about the rapidly evolving technology and adoption curve, Gary Tyreman, CEO of Univa, said in an interview Tuesday.
"We could have done it but it would have been fairly lightweight," he said of a March 2015 release. "It wouldn't have been very meaningful. We have validated with end users, as well as the ecosystem itself, what the key challenges are when you put Docker in the enterprise."
These hurdles typically revolve around orchestration and scheduling, as well as management and configuration, especially as enterprises' stable of Docker containers grow. Grid Engine Container Edition includes cluster-wide full job control with pre-emption, termination, and run-time statistics, the developer said. The cluster works with any operating system or infrastructure, as well as thousands of applications and frameworks, according to Univa, allowing IT to maximize Docker resources, gain velocity, and schedule Docker containers for appropriate prioritization and usage.
Without the correct optimization and management tool in place, enterprises risk losing the benefits of containerization due to more configuration challenges and deployment headaches.
In July, Univa became a founding member of the Cloud Native Computing Foundation (CNCF), a community of vendors on a mission to "create and drive a new computing paradigm that is optimized for modern distributed environments," including container-packaged technologies. The developer also signed on signed on with the Google Cloud Platform Partner Program as an Authorized Technology Partner, evidenced by its integration between Univa's solutions suite and the Google Cloud.
These moves are all part of Univa's process of parlaying its success with existing customers to a broader base of enterprises, said Tyreman.
"We believe we can snap in for orchestration. It's been battle-tested for years, in the enterprise, in a messy – what I mean is heterogeneous – environment of Windows, mixed Linux, cloud," he said. "When you have as many containers – a factor of n – compared to virtual machines, that gets to be pretty complex, very fast."
While other developers target container orchestration, Univa's expertise in large organizations should place it in good stead with prospective customers, Tyreman said.
"I think the construct of some of the orchestration tools out there expect one container per second. We have been capable of doing many thousands of placements per second. Our work with the CNCF, what we're interested in contributing, is anything that helps scale. And scale is not 100 nodes," he noted. "Our customers sit with 5,000, 6,000, 20,000 nodes. Micro services, which are ideal for containers, come and go. Instead of 100 apps running, you could have millions of containers running. Latency is going to be a critical factor in all of this. That's our sweet spot."
That was the case for the Centre for Genomic Regulation (CRG), which used Docker containers with Univa Grid Engine for scientific data analysis on distributed clusters. "This solution allowed us to reduce configuration and deployment problems and rapidly create a self-contained run-time environment to accelerate development," said Paolo Di Tommaso, CRG software engineer, in a statement.