The videos from the Summit keynotes are now available! 

Event Details

Please note: All times listed below are in Central Time Zone

<< Go back
Real Telco OpenStack upgrade lessons learned
Telecom & NFV

Queens release was the first one which addressed OpenStack problem of upgrades between relatively frequent releases.

Fast Forward Upgrade (FFU), Skip Upgrade or (N+3) upgrade are the terms describe the same thing, upgrade from OpenStack Newton to OpenStack Queens.

We are now a year after Queens release and it may be a good time to look again on the N+3 upgrade,
when Telcos are scheduling to perform upgrades on their own infrastructure and understand the complexity of these upgrades.

In Telco world, OpenStack upgrade can't be isolated from other components and triggers other elements upgrade:

  • operating system upgrade
  • kernel drivers upgrades
  • HW firmware upgrades
  • and finally, VNFs workload running on top of the cluster which often need to be upgraded as well

On this session, I am going to discuss the most common problems we've seen so far?
Share some lessons learned. And compare TripleO (N+3) upgrade with other ways of doing the upgrades.


What can I expect to learn?

OpenStack upgrades are difficult and (N+3) Fast Forward Upgrade is even more challenging. This is a complex procedure which is hard to fully automate as it's rather unreliable and depend on customer-specific configuration.

There are multiple projects out there which shows how to do an upgrade in a few minutes, but is it really a full upgrade? Docker images swap is NOT a full upgrade.
The full upgrade includes also upgrade of underlying infrastructure:

  • VNFs workload upgrade
  • Ceph upgrade
  • operating system upgrade
  • kernel drivers upgrades
  • firmware upgrades

Full cluster reboot after upgrade with VNFs running on top of OpenStack cluster is slightly more complex than a reboot of a cluster running web type workload.

Ceph upgrade is a separate and complex beast on its own especially when you want to migrate Ceph data from FileStore to Blue store during your upgrade.

There is a lot of corner cases which you should know and be prepared before your upgrade. The main goal of that session is to share experiences we gained doing real production OpenStack upgrades and FFU with TripleO.

Tuesday, April 30, 11:40am-12:20pm (5:40pm - 6:20pm UTC)
The Colorado Convention Center - Meeting Room Level - 505/506
Slides: Real Telco OpenStack upgrade lessons learned
View video
Difficulty Level: Intermediate
Tags: Technical Case Study Arch / Ops OpenStack Ceph TripleO
Radosław Śmigielski
Citrix
Radoslaw has started virtualization advanture with opensource Xen in 2007. He has joned Citrix where Xen and virtualization adventure became proffesional in form of XenServer product. Also at Citrix started next big, cloud adventure with CloudStack, if anyone still remembers cloud.com. In 2013, he has jumped into  OpenStack world at Aclatel-Lucent (later acquired by Nokia) where he works and... FULL PROFILE
Gabor Szitas
Mr.
Gábor started his career at GE Healthcare as an intern (C++) which led him to join Nokia (at that time Nokia Siemens Networks) in 2013. During the upcoming years, he's worked on the NT-HLR development and built up his knowledge about Telco Core networks and Hardware Security Modules. The "Cloud Era" reached him in 2015 when Nokia started to invest OpenStack based projects. There he worked... FULL PROFILE