BOSTON — On the third day of the OpenStack Summit 2017 this week, there was a session entitled “What is OpenStack?” I found that kind of curious. You would think most people attending an OpenStack Summit would have some idea what it is.
Yes, I attended the session. But I have a good excuse. I’m a journalist who admittedly doesn’t work with OpenStack software, and I was just looking for somebody to speak in plain language that I could understand.
So, imagine my surprise when I arrived at the “What is OpenStack?” session to find that it was standing-room only.
The presenter of the session, Sander van Vugt, was surprised as well. “I thought maybe two people would show up for this,” he said. “It is an OpenStack conference after all.”
Van Vugt, who works as an independent Linux author, trainer, and consultant, did a great job of giving the packed room a rudimentary introduction to OpenStack.
The OpenStack project formed in 2010 as an open source group to write software for a cloud operating system. It controls large pools of compute, storage, and networking resources throughout a datacenter, all managed through a dashboard. OpenStack began as a joint project of Rackspace and NASA. As of 2016, it is managed by the OpenStack Foundation. More than 500 companies have joined the project.
It has separate sub-projects working in nine areas: compute, storage, networking, analytics, security, management, deployment tools, application services, and monitoring.
Van Vugt highlighted some of the most productive and popular projects within these sub-groups.
The Nova group, within the compute sub-project, was one of the first two original projects in OpenStack. It focuses on hypervisors. “Infrastructure-as-a-Service was realized through Nova,” said van Vugt.
The Glance group, also within the compute sub-project, focuses on images. “The challenge in cloud is you want to deploy instances but not install instances,” said van Vugt. “You deploy instances as simply as possible via Glance.”
The Swift group works on object storage, while the Cinder group looks at block storage. “You don’t want your storage to be bound by one physical machine,” said van Vugt. “If Cinder block storage is only available locally on hypervisor nodes, that’s not very scalable. That’s why we need Swift.” But he added that Ceph software also does object storage. Even though Ceph is not formally a part of OpenStack, most OpenStack deployments are running Ceph, rather than Swift, he said.
The Neutron group deals with software-defined networking (SDN). And the Heat group deals with orchestration of virtual machines (VMs).
DistrosOpenStack cloud users can grab the freely-available code and start working with it. However, hardly anyone actually does that, apparently. Most users work with OpenStack distributions put out by various vendors. “There are quite a few OpenStack distributions,” said van Vugt. “They will make things easier for you.”
While cloud OpenStack updates its code every six months, most distributions (distros) get updated every 12 to 18 months. The most popular distros come from Red Hat, Canonical, Mirantis, and Cisco, according to the most recent OpenStack User Survey.
VMware is gaining some traction with its distro — VMware Integrated OpenStack (VIO). Customers that purchase VMware’s vSphere receive the rights to VIO as part of that offering.
“There’s only a small amount of awareness of VMware and OpenStack,” said van Vugt. “What I consider interesting is that VMware’s OpenStack runs on top of vSphere; their NSX is your SDN; and their vSAN is your object storage. They did that in a very clever way. If you’re already using vSphere it makes it very easy to get OpenStack on top.”
Finally, for those IT people shopping for a cloud OpenStack distribution, van Vugt recommended checking the following criteria:
- Supported hypervisors
- Availability and licenses
- Requirements for getting started
- How it connects to current infrastructure
- The support offering