pixel - Fotolia
Portworx revised its software to boost data management and storage provisioning in containerized applications.
Portworx Enterprise 2.6 improves operational performance for clustered Kubernetes nodes. It extends the startup's effort to help data centers provision and manage persistent storage for stateful cloud applications.
Enhancements cover optional PX-Backup and PX-Autopilot for Capacity Management modules, which are licensed separately in Portworx Enterprise.
The Portworx storage software shares the Kubernetes control plane. The vendor claims this sidesteps performance issues associated with node failure in external networked storage. Customers have the option to run Portworx as software-defined storage on commodity servers, but it could also expose volumes running on a SAN.
The privately held vendor does not disclose revenues, but claims its recurring license revenue jumped 61% over the last three quarters. Portworx customers include a division of carmaker Audi AG, Comcast, GE Digital, Lufthansa, T-Mobile, and Ford Motor Co.
Balanced storage performance for Kubernetes
Although containers consume less storage overhead than virtual machines (VMs), IT administrators still must keep tabs on containers to prevent stranded resources. To that end, the Portworx Enterprise platform added node-capacity rebalancing to optimize existing storage. No single node is fully allocated, as long as other nodes have storage capacity to accept new replicas. Portworx tells Kubernetes which nodes are available for data placement.
"We make sure you fully utilize your existing storage before you bring on new nodes," said Michael Ferranti, a Portworx vice president of product.
Kubernetes uses a key value store, known as etcd, to retain metadata about clustered nodes and control planes. Ferranti said Portworx Enterprise 2.6 beefs up system reliability to ensure a Portworx cluster remains operational in the event of an etcd outage.
Other new features include support for K3s, a lightweight distribution of Kubernetes geared to edge computing environments, and support for proxy volumes.
PX-Backup 1.1 is in preview now. The telemetry helps customers assess system state of their container architecture. Granular restores attack a common problem with Kubernetes: the necessity to back up application data and application configuration in its entirety. PX-Backup captures all that information as one unit, allowing users to restore a single piece of configuration data.
The backup extends to custom objects that users create and add to a Kubernetes cluster. Portworx said PX-Backup now allows quotas to be set at the namespace level for CPU, memory and storage, with backup metrics and the ability to establish default backup schedules per application.
Florian Buchmeier, a DevOps engineer at Audi Business Innovation (ABI) in Munich, said his teams started using containers several years ago to replace a Jenkins-based continuous integration platform running VMs in Amazon Web Services.
Buchmeier said ABI selected Portworx storage after determining it was easier to manage than scale-out file storage and more scalable than legacy NAS. Along with mobility services for vehicles, ABI developers provide project management and documentation services, managing the projects with Atlassian Jira instances and Bitbucket storage. Persistent Portworx storage on the back end provides read/write volume targets for multiple front-end Kubernetes applications.
"We looked at the storage providers that supported Kubernetes' ReadWriteMany volume access, but who also excelled at reliability. We were very keen on performance, support and pricing. Portworx checked all the boxes for us," Buchmeier said.
Portworx introduced PX-Autopilot last year, allowing users to add storage incrementally to a cluster based on consumption. Version 1.3 adds features around GitOps configuration management and managing node pools with Google Container Engine.
Portworx Enterprise 2.6 is slated for general availability (GA) on Tuesday, and PX-Autopilot goes GA on Aug. 31.