Every six months or so, the OpenStack community seems to swing between euphoria and despair – depending on the amount of progress being made in deploying OpenStack in production environments.
Despite a well deserved reputation for being challenging to implement, enthusiasm for running all kinds of application workloads on OpenStack was running high at the OpenStack Summit Austin 2016 conference, thanks largely to advances in tooling that make it simpler to deploy and manage OpenStack.
For example, Intel this week demonstrated at the conference an open source Cloud Integrated Advanced Orchestrator (CIAO) tool that combines three separate OpenStack schedulers into one unified scheduler. Imad Sousou, vice president and general manager for open source software at Intel, says Intel has used CIAO to launch 10,000 Docker containers and 5,000 Fedora virtual machines in the same 100-node cluster in just one minute. In addition, Sousou says CIAO also provides unified networking out of the box between virtual machines and containers.
Also at the Summit, AT&T revealed that, thanks to investments it has made in automation, the amount of time required to deploy OpenStack clouds has been sharply reduced. Coho Data, meanwhile, announced it has integrated the OpenStack Cinder API for storage in its systems, along with support for a software-defined workload distribution platform that supports both OpenStack and VMware.
Those advances come on the heels of a Mirantis move to extend the number of turnkey appliances it offers to simplify OpenStack deployments.
As OpenStack becomes simpler for IT organizations to cope with, there’s been a marked increase in the number of production deployments. A full 65 percent of 318 IT professionals surveyed by the OpenStack Foundation report they are now using OpenStack in production. While the majority of the organizations using OpenStack tend to have access to extensive internal engineering resources, more organizations are clearly starting to standardize on OpenStack to primarily run “cloud-native” application workloads.
For example, Mario Müller, vice president of IT Infrastructure for the Volkswagen Group, told Summit attendees that the automotive manufacturer expects to have OpenStack running in a production environment by the middle of this year. It will be deployed in a 2,000 square-meter data center that Volkswagen created to host new applications that will run on an OpenStack distribution curated by Mirantis. Müller told conference attendees that OpenStack is core to “the reinvention of IT infrastructure” at Volkswagen.
Obviously, not every IT organization has the resources of Volkswagen. But with each successive update of OpenStack, progress toward making the framework easier to employ is being made at a rapid pace. As a result, a recent survey of 650 IT professionals conducted by Talligent (a provider of tools for tracking cloud computing costs) and VMblog found that the majority expect to one day run both cloud-native and enterprise legacy applications on OpenStack. It may be a while before those traditional legacy applications get deployed on OpenStack, but even SAP revealed at the OpenStack Summit that it is standardizing delivery of its cloud application portfolio on OpenStack.
For now, however, most of the OpenStack community is wrestling with upgrading to Mitaka, the 13th release of OpenStack. The major benefit of this version of OpenStack is that a number of previously manual processes have been streamlined. But it’s also clear that there are still OpenStack interoperability issues that need to be addressed, and the framework itself is still a long way from being something that could be mastered by the average IT administrator. Nevertheless, the momentum behind OpenStack, at least for the moment, continues to build.