CERN is rapidly expanding OpenStack cores in production as it accelerates work on understanding the mysteries of the universe.
The European Organization for Nuclear Research currently has over 190,000 cores in production and plans to add another 100,000 in the next six months, says Spyros Trigazis, adding that about 90 percent of CERN’s compute resources are now delivered on OpenStack.
RDO’s Rich Bowen shot the video, which runs through CERN’s three-and-a-half years of OpenStack in production as well as what’s next for the humans in the CERN loop — the OpenStack team, procurement and software management and LinuxSoft, Ceph and DBoD teams.
Trigazis also outlined the container infrastructure, which uses OpenStack Magnum to treat container orchestration engines (COEs) as first-class resources. Since Q4 2016, CERN has been in production with Magnum providing support for Docker Swarm, Kubernetes and Mesos a well as storage drivers for (CERN-speciﬁc) EOS and CernVM File System (CVMFs). Trigazis says that many users are interested in containers and usage has been ramping up around GitLab continuous integration, Jupyter/Swan and FTS. CERN is currently using the Newton release, with “cherry-picks,” he adds.
— Unix (@UNIXSA) February 3, 2017
Upcoming services include baremetal with Ironic; the API server and conductor are already deployed and the first node is to come this month. Another is workflow service Mistral used to simplify operations, create users and clean up resources. It’s already deployed and right now the team is testing prototype workflows. FileShare service Manila, which has been in pilot mode since Q4 of 2016, will be used to share configuration and certificates.
For updates from the CERN cloud team, check out the OpenStack in Production blog.