When I hear answers to the question, "What is OpenStack?" I hear echoes of Morpheus trying to explain to Neo, "What is the Matrix?" Morpheus can only talk in metaphors. The Matrix is a dream world, a battery, control. All are true, but none quite captures what it really is.
OpenStack likewise is defined in metaphors. In the past year I've heard proponents describe OpenStack as "people," an "interface," a "set of APIs" and a "platform." Last week at the OpenStack Summit in Austin, we also heard it be described as a "strategy for taking advantage of diversity in IT," an "integration engine," and, according to CoreOS CEO Alex Polvi, OpenStack is simply "an application."
All are true, of course, which is part of the allure of OpenStack and also the source of its ongoing identity crisis. OpenStack is six years and 13 versions into its life as an "open-source software platform for cloud computing," which is how Wikipedia puts it and actually is a pretty good definition.
I'll throw another metaphor into the pot: OpenStack is a use case. There are a lot of them, and that's part of the identity problem: Which use case? According to the annual OpenStack User Survey, the number one use case is test and development (63 percent), which is not surprising because that's where all cloud efforts begin. Others include infrastructure as a service (49 percent) and Web services and e-commerce (38 percent). The total percentages exceed 100 because survey respondents were allowed multiple choices.
The most intriguing use case, however, is network-functions virtualization (29 percent). Work has been quietly proceeding on NFV for the past few years and no one really noticed until AT&T and Verizon recently announced large network architectures deployed on OpenStack.
AT&T claims to be the biggest OpenStack implementation after building 74 AT&T Integrated Cloud (AIC) sites with another 30 coming this year as the telecom service provider works toward its goal of virtualizing 75 of its network operations by 2020.
AT&T didn't wait for analysts or the media or anyone else to declare that OpenStack was ready for production. They did it, as one official told me earlier this year, simply because "OpenStack is a great platform on which to deploy a network."
They came to that decision not because the company wanted to become a cloud computing leader, but because of the need for telcos to become more open and agile or risk not being able to keep up with the demand for network services. The result for AT&T was ECOMP (Enhanced Global Control Orchestration Management Policy), which Sorabh Saxena, AT&T's senior vice president for software development and engineering, described in detail to Summit attendees.
Boris Renski, co-founder and chief marketing officer of Mirantis, whose OpenStack distribution AT&T uses in the AIC, explained how OpenStack also enables telcos to keep their costs under control as they build out enough data centers to meet demand.