Portworx, a stateful Docker container storage and data services provider, was recently named a 2017 Cool Vendor of Storage Technology by the Gartner Group. InfoQ took a look at this technology, and we thought it was indeed rather cool!
The Portworx PX-Series addresses the five most common DevOps problems: persistence, high availability, automation, security, and infrastructure, offering a variety of features that allow DevOps to move from development to production. Use cases supported by Portworx include:
- Databases, relational and NoSQL
- CI/CD, such as Jenkins and Bamboo
- Big Data, such as Hadoop and Spark
- Content Management, such as Word Press, Drupal, and Joomla
- DC/OS, such as Mesosphere, Elastic, Hadoop, and Kafka
Recently released version 1.2 introduced “Cloud Snap”, a utility that “enables users to take a full or differential copy of a Portworx container volume and automatically upload it to their cloud provider of choice.” Portworx can be installed on top of Kubernetes, Mesosphere, Docker, and Rancher.
According to Portworx:
Portworx is the solution for stateful containers, designed for DevOps. With Portworx, users can manage any database or stateful service on any infrastructure using any container scheduler, including Kubernetes, Mesosphere DC/OS, and Docker Swarm. Portworx solves the five most common problems DevOps teams encounter when running containerized databases and other stateful services in production: persistence, high availability, data automation, support for multiple data stores and infrastructure, and security.
Eric Han, VP of product management at Portworx and a founding member of Google Kubernetes, spoke to InfoQ about the hyperconverged container market, where it may be heading, and the problems DevOps face running containerized databases.
InfoQ: How long have you been with Portworx and what are your current responsibilities?
Eric Han: I came to Portworx two years ago from Kubernetes after the release of version 1.0. True back then and now, the ecosystem is moving fast.
Customers might be trying multiple components and can be rest assured that for any scheduler or cloud environment, Portworx enables stateful containers. Customers trust Portworx to do two things: (1) enable stateful containers, and (2) never lose data in this distributed compute world. A lot of my work has been spending time with customers and seeing how we can help them enable their move to containers or the cloud. A lot of that energy renders out into our product roadmap, where my most important responsibility is judiciously saying ‘no’ to features that we could do, but shouldn’t.
InfoQ: Are there any issues moving from development to production especially in terms of performance?
Han: Performance is not an issue when moving into production, as it’s almost always proven in pre-production tests. However, customers also want to prove seamless failover to their satisfaction, which sometimes is subject to their imagination.
We do a lot of battle testing. We work with customers on the standard tests, but some push strongly for extreme tests such as push cascading failovers. This is where we've benefitted from being a first mover; lots of energy from customers that benefits all users.
InfoQ: What makes Portworx unique over competing hyperconvered container platforms especially in terms of the five most common DevOps problems?
Han: What separates us is three aspects:
- Production proven: Other platforms might allow deployment on day zero, but then most struggle with management on day two and after. We’re in more production deployments than any other vendor, and that experience has translated to Portworx having a much greater breadth of services.
- Scheduler integration: Anyone wanting to run a cloud native application needs to be able to manage their data lifecycle in a coordinated manner with their compute scheduler. We’re the only one that can be deployed by a scheduler, automatically coordinate volume to container placement, and that scales to thousands of nodes.
- Breadth of services: We are the only solution that provides protection in the form of replication, multi-cloud backup, and encryption. These are all configurable features but the mix let's users move across environments with any workload.
InfoQ: How do you see the hyperconverged container platform market evolving in, say, the next five years?
Han: Again, deploying stateful containers with virtual-machine-based infrastructure will seriously hinder running stateful containers in production. We think in the next three years, we’ll see vendors that help organizations move to the cloud, be effective at hybrid, and go beyond. We’re excited by more pure players entering the market.
InfoQ: What’s on the horizon for Portworx?
Han: Reflecting on the past year, our product gave customers a way to run stateful containers effectively regardless of their environment: cloud, bare metal, or in virtual machines. We abstracted storage into a cluster. We have already started building multi-cluster capabilities that lead us towards being the first to do disaster recovery with stateful containers. We’ll show you what that leads to in our version 2 release slated for the end of this year.