Singularity Moves Up the Container Value Chain
The enterprise version of the Singularity HPC container platform released this week by Sylabs is designed to allow users to create, secure and share the high-end containers in self-hosted production deployments.
The enterprise version emerged from beta testing that commenced in April. The upgraded edition of incorporates a “remote builder” service that allows developers to use Singularity’s image format without restricting access on shared resources. Instead, the resulting files can be signed to verify authenticity when the files are used.
The service allows users to build containers for ARM, Power and architectures other than x86. It can also be used in conjunction with continuous integration/continuous delivery pipelines to automate deployment, the company said.
Also included is a container library which allows the sharing of image files while also tracking file versions to meet audit requirements. The library is promoted as a way to capture and reproduce “corporate memory” by separating applications from underlying tools. That capability, Sylabs said, eliminates dependency on a particular Linux kernel or distribution.
Those security features are backed by a repository of certificates, or keystore, that maintains a “chain of trust” persisting throughout the lifecycle of Singularity’s image files.
Early adopters of Singularity “overwhelmingly seek services that ensure a seamless transition from proof-of-concepts to production,” said Gregory Kurtzer, Sylabs’ CEO and founder. Kurtzer spearheaded development of Singularity while at Lawrence Berkeley National Lab.
The self-hosting feature is intended to streamline deployment either in datacenters, private clouds or in hybrid cloud configurations, Kurtzer added.
The upgrades address lingering challenges association with shifting application containers from development and testing to production. While a recent Stack Overflow survey found growing adoption of containers among about 90,000 software developers polled, production deployments remain low.
“It is clear that the transition to production is fraught with challenges,” the company notes.
Sylabs, Albany, Calif., also recently announced support for the Mac operation system with the release of a desktop version of Singularity. The beta version allows Linux-based containers to be built, tested and verified on the Mac OS.
The upgrades reflect the steady embrace of what Sylabs calls EPC, for “enterprise performance computing.” The startup’s strategy involves moving beyond services to handle more demanding artificial intelligence, machine and deep learning as well as advanced analytics workloads.
--Editor's note: The story has been updated to clarify the processor architectures supported by the Singularity platform.