When OpenStack launched in 2010, there were only two projects at the time: Nova compute and Swift storage. Over the last seven years, OpenStack has gotten significantly larger and more complicated, with many different projects that are all part of the open-source cloud platform effort.
In a session at the OpenStack Summit in Sydney, Australia on Nov. 8, Thierry Carrez, VP of Engineering at the OpenStack Foundation, detailed a new effort to help map the OpenStack landscape.
"We can't communicate really well by trying to put everything on a map," Carrez said.
Having too many things on a map makes it too crowded and often confusing, which is something that Carrez showed with multiple historical maps. Looking at OpenStack, Carrez said the official maps from the OpenStack Foundation have gone through several generations.
One of the early maps Carrez showed included the different compute and storage projects and connected them with common networking, though in his view that early map was still missing things and was very confusing.
Earlier this year, another attempt was made at mapping OpenStack projects by assigning icons to each project and then simply listing them all out, which Carrez said was also very confusing.
"The message you get out of it is there are just too many things in OpenStack," Carrez said. "The mistake with that approach is it shows all the project teams, but that doesn't match with what users want to actually install."
Carrez emphasized users don't need to know every single project and plugin in order to benefit from OpenStack. As such, a new approach for mapping OpenStack was determined, with several key requirements.
Among the requirements for the map was to show official OpenStack components and not try to provide a map of the whole OpenStack ecosystem. Carrez said another requirement for the map is that it is consumer-centric, showing deliverables rather than product teams. Additionally, it is aligned around buckets, based on who consumes the functionality.
Applying the Art of Cartography to Mapping OpenStack
"What I learned from the art of cartography is that it's about what you remove as much as what you put in the map," Carrez said. "The map is meant to help people make choices and see the options within OpenStack."
The right side of the map (shown below) includes items that are used to help facilitate the operations of an OpenStack cloud. At the bottom of the map are projects to look into to help manage the lifecycle of an OpenStack cloud.
"The map makes clear a few core areas, including compute, networking and storage that live on top of a number of shared services," Carrez said. "It's not about dependencies; the map is meant to represent the things OpenStack produces that a user might be interested in deploying."
Among the items on the map that has been debated within the OpenStack community is the placement of the Zun container project. The Zun container project got started at the end of 2016. On the new OpenStack map, Zun now holds a spot in the compute layer, alongside the core Nova compute service and the Ironic bare-metal project.
Carrez said the new OpenStack map will be updated and versioned regularly to help properly define OpenStack for users. He also noted the goal isn't to communicate everything with a single map and that additional maps will be developed to communicate other use-cases as needed.
Sean Michael Kerner is a senior editor at ServerWatch and InternetNews.com. Follow him on Twitter @TechJournalist.