Sylabs has updated the enterprise edition of its Singularity container platform to include the updates to the open source platform that were made last fall.
Company CEO Gregory Kurtzer says key capabilities now included in SingularityPRO supported by Sylabs includes a plug-in architecture that makes it easier to extend the functionality of the company’s container platform in addition to the Open Container Initiative (OCI) and Container Networking Interface (CNI) developed under the auspices of the Cloud Native Computing Foundation. OCI support is critical because it allows container images to run on both Singularity and Docker container platforms. Singularity is also compatible with container orchestration engines such Kubernetes and Nomad from HashiCorp.
Singularity is a container platform originally employed primarily by organizations running high-performance computing (HPC) applications. But as more organizations have started to build artificial intelligence (AI) applications using containers, Kurtzer says Singularity is increasingly finding its way into enterprise IT environments that want support from Sylabs. The most recent edition of Singularity also added support for encryption via a key management capability that is now generally available.
Singularity primarily differs from Docker in that it is based on a single-file format that includes the runtime environment. Kurtzer says that approach makes it easier to deploy and support because the need for “tar balls” of files has been eliminated. In fact, he notes that Singularity not only can be deployed on everything from x86 servers to graphical processor units (GPUs) and field programmable gate arrays (FPGAs), but it also runs natively on MacOS. The Singularity approach also provides better security because there’s no need to grant users control of a root-owned daemon process or kernel feature, he adds.
Kurtzer says there are now more than 25,000 users deploying Singularlity containers that are supported via the enterprise edition of the platform. There are, of course, thousands of other users that Sylabs might not be aware of, given the open source nature of the platform. Kurtzer notes the company is working on a 3.2 update to the enterprise edition that it plans to deliver next month.
In the meantime, Sylabs is making a case for a second container format that lends itself better to supporting microservices-based applications at scale. Singularity may not have the same level of momentum as Docker. But Sylabs did recently sign agreement with both Google and Microsoft to make it easier to deploy Singularity containers on their respective public clouds. Lenovo, meanwhile, has committed to making Singularity part of the container orchestration platforms it supports on servers deployed in on-premises IT environments.
It remains to be seen to what degree IT organizations will embrace multiple types containers engines. But as microservices continue to evolve, there does appear to be a high-end niche that Docker containers don’t address as well as Singularity. As enterprise applications based on microservices become more complex, many of them may consider alternative approaches. Regardless of the path chosen, it’s clear containers running on everything from virtual machines to bare-metal servers will soon be everywhere.