This event has ended. Visit the official site or create your own event on Sched.

Log in to bookmark your favorites and sync them to your phone or calendar.

Operations & Admin Track [clear filter]
Thursday, October 19


OpenStack maintenance - Finding the needle in the haystack
Many services, nodes, configuration files, log files - due to this complexity it can be cumbersome to uniquely identify the root cause for problems in a timely manner. We’re going to look at common problems in OpenStack environments, their root cause, and options for efficient operation.

OpenStack maintenance - Finding the needle in the haystack

OpenStack has come to stay for good. Many companies leverage the advantages of having a private cloud: fewer costs, more control, no vendor lock-in. In addition, you get a better understanding of how cloud platforms work. However, you also need to take care of operations and maintenance. OpenStack troubleshooting is a nontrivial task and takes a lot time, knowledge, and experience. Hundreds of log files are written by numerous services with several configuration files to countless virtual and physical machines - the possibilities for errors seem endless. Manual root cause analysis is like looking for the needle in the haystack. We’re going to look at common problems in OpenStack environments, analyze their root cause, and discuss options for effective and efficient operation and troubleshooting.

Participants of this session will learn
  • Basically, OpenStack is a micro-service application and therefore also needs to be monitored
  • Insight into what is important to monitor in an OpenStack environment – Resource utilization of physical and virtual machines – Health of supporting services (MySQL, RabbitMQ, Memcached) – OpenStack service availability – Response times of OpenStack services – Lead times of common OpenStack operations
  • A brief overview of monitoring tools for OpenStack
  • Common error patterns in OpenStack environments and how to identify and fix them

avatar for Dirk Wallerstorfer

Dirk Wallerstorfer

Director Technology Strategy, Dynatrace
Dirk Wallerstorfer is Director of Innovation Strategy at Dynatrace. He has 15+ years of hands-on experience in networking, security, and software engineering. Dirk's job is to know everything that is going on around virtualization of infrastructure and application environments. He... Read More →

Thursday October 19, 2017 13:40 - 14:10


Troubles and Tribulations Running Production OpenStack Environments
Running OpenStack in commercial production environment is a non-trivial task. Running it at scale, in geographically distributed environments is quite hard. We’ll share some hard fought experiences from running our production OpenStack environment since the last 8 years.

City Network AB together with OP5 AB will share some insights in the daily challenges of planning, deploying, upgrading and servicing a geographically distributed, multi-data center OpenStack production environment. We’ll go over a few specific problems we’ve had, how we responded and what we learned, the post-mortem we conducted and what we could’ve done better. We’ll also talk about what tools we use to support, monitor and trouble-shoot our environment and what we’d like to see.

We’ll also talk about where we see our environment going forward and the challenges to monitor, visualise and react to streaming metrics environments at scale. We’ll touch upon on how OpenStack stacks up from a traditional sys-admin perspective.

This session will be a good blend of high level information, some hard-fought lessons from the field but also a good pointers about where we see production OpenStack deployments going in the future.

avatar for Yavor Adel

Yavor Adel

PTL NextGen Architecture, OP5
avatar for Daniel Öhberg

Daniel Öhberg

VP of Operations, City Network Hosting AB
Responsible for IT Operations and development at City Network. Background in IT Operations and company management. Fairly new to Openstack, but a long record in IT infrastructure and virtualization.

Thursday October 19, 2017 14:20 - 14:50


Building a public cloud service on OpenStack
A talk filled with war stories from designing and building a multi site public cloud service on OpenStack with all the bells and whistles expected, including a custom control panel. And keeping the beast running and up to date.

  • Thoughts on design - going with community or propriatary vendor solutions.
  • Due to hysterical reasons - from manual installation to Ansible Openstack and the path between
  • A bewildering story of networking.
  • Handling hardware
  • Around the world in 80 ms - coping with multiple locations and making them cooperate.
  • The sad state of billing.
  • Minor bugs, issues and and live patching.
  • Vendor support - or the lack thereof.
  • The API and coding against it.
  • The importance of test and development systems.
  • Supporting the mess, from customer input to fix in production.
  • Keeping the beast up to date.
  • Some final advices.


Magnus Bergman

Senior Network Engineer, City Network Hosting AB

Thursday October 19, 2017 15:20 - 15:50
Carstensen 1 floor below the Foyer


Connecting Kubernetes and OpenStack with simple/scalable L3 networking and Network Policy

As OpenStack embraces Kubernetes as the container platform of choice, it becomes imperative to connect VM’s and containers seamlessly.

This talk outlines Project Calico’s pure-L3 network approach combined with declarative (Kubernetes) Network Policy to interconnect OpenStack with Kubernetes.

As the OpenStack community embraces Kubernetes as the container orchestration platform of choice, it becomes imperative to seamlessly connect applications across VM’s and containers.

Early OpenStack Neutron plugin implementations approached network isolation by creating network overlays on top of convoluted paths through vswitches, bridges and L3-agents on compute nodes, along with coarse security groups. In contrast, Kubernetes and other container orchestrators provide rich network policy constructs while assuming IP connectivity between nodes and containers/pods, facilitating simple, scalable network design and isolation with network policy.

This talk explores a pure-L3 network approach combined with declarative (Kubernetes) Network Policy, espoused by Project Calico, to interconnect OpenStack with Kubernetes across 3 scenario’s: 1. Kubernetes alongside OpenStack (with application instances in either) 2. Kubernetes running within OpenStack VM’s 3. OpenStack as a (containerized) application orchestrated by Kubernetes


Karthik Prabhakar

Director of Solution Architecture, Tigera
After years of experience with large cloud infrastructure deployments involving Kubernetes or OpenStack, Karthik discovered the power of Project Calico, combining a scalable, simple IP routed network fabric with distributed intent-based policy enforcement designed for cloud-native... Read More →

Thursday October 19, 2017 16:00 - 16:30
Carstensen 1 floor below the Foyer