CNCF Adopts Longhorn Storage Project from Rancher Labs
The Cloud Native Computing Foundation (CNCF) announced this week that a Longhorn block-based storage software for Kubernetes clusters developed by Rancher Labs has become an open source CNCF sandbox project.
Rancher Labs CEO Sheng Liang says Longhorn will make it significantly easier to deploy stateful applications on top of Kubernetes clusters by eliminating the need for specialized drivers.
Version 0.6.2 of Longhorn includes support for volume snapshots, built-in backup and restore, live upgrades without impacting running volumes, cross-cluster disaster recovery with defined recovery time objectives (RTO) and recovery point objectives (RPO), one-click installation and a graphical user interface.
Based on 30,000 lines of Go code employed to create separate engine and management plane, Longhorn is significantly lighter than traditional storage software because it builds on existing Linux storage primitives, Liang says. It also doesn’t require a dedicated storage administrator to deploy and manage. It’s designed from the ground up to be used by the same team managing the Kubernetes cluster, he notes, adding modern storage hardware such as solid-state drives (SSDs) and NVMe backplanes made it easier to build Longhorn without compromising performance.
While some applications might want to access block storage directly, Liang says Rancher Labs expects most organizations will layer a file system on top of Longhorn to access various forms of persistent storage.
Getting Longhorn to this point has required significant effort on the part of Rancher Labs because the underlying programming tools and storage technologies kept evolving rapidly, says Liang, noting a team of developers from multiple vendors will be needed to take Longhorn to the next 1.0 level.
One of the most appealing aspects of Longhorn is that it doesn’t require IT teams to acquire separate snapshots and disaster recovery applications on top of the core storage engine; because of that, Longhorn could significantly reduce the total cost of storage compared to legacy IT environments.
While the number of stateful applications being deployed on Kubernetes is steadily increasing, there’s still a need for an easier means of accessing persistent storage from within a Kubernetes cluster. Longhorn represents the most comprehensive approach the CNCF has chosen to pursue to date.
It’s not clear how many storage vendors will embrace Longhorn should it graduate to become a fully supported CNCF project. However, at a time when the management of compute and storage already is starting to converge, that may not matter all that much. Providers of distributions of Kubernetes one day soon may simply bundle Longhorn with the rest of the add-on modules for Kubernetes they already provide to access commodity storage.
In the meantime, stateful applications that require access to a database or some other form of persistent data store will continue to be deployed at ever faster rates. The developers of those applications can now also take some comfort in the fact that, thanks to the CNCF, the pain associated with accessing the data needed to drive those applications soon might be substantially less than it is today.