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

CoreOS Fest Berlin, the Open Source Distributed Systems Conference, brings together experts from all over the world to discuss container technologies, distributed systems and open source software to help take the industry to the next level in distributed computing.

The 2016 event will gather 500+ developers, devops professionals and sysadmins together to explore the next steps in making container technologies successful and secure.

Learn more at coreos.com/fest 

Sign up or log in to bookmark your favorites and sync them to your phone or calendar.

CoreOS [clear filter]
Monday, May 9

09:50 CEST

etcd over gossip: strongly-consistent semantics in a weakly-consistent world
etcd is a powerful and robust core component of modern infrastructures. Weave Mesh is the gossip protocol at the core of Weave Net, the battle-tested Docker SDN. This talk chronicles the implementation of etcd-on-Mesh, hosting the strongly-consistent etcd API on an eventually-consistent gossip protocol. The project enables developers to use etcd in a much wider variety of contexts than is currently assumed. For example, embedding an etcd cluster in their application and get an ephemeral lock store, or deploying a permanent etcd cluster in a heterogeneous, partially-connected network environment.

In this talk, I describe the core semantics of the Raft protocol, the requirements it has for peer-to-peer communication, and how it might be usable over a packet-oriented, rather than stream-oriented, transport. (Think: UDP rather than TCP.) Then, I describe the basics of gossip protocols, how they model state transfer, and Weave Mesh's particular approach. Finally, I describe how these two models of distribution can be married together, and how each one accommodates and compensates for deficiencies in the other.

avatar for Peter Bourgon

Peter Bourgon

Typist, Weaveworks
Peter Bourgon is a distributed systems and infrastructure engineer who has seen things. He works at Weaveworks, contributing to the Weave Scope monitoring tool, and to Weave Mesh, a gossip layer for distributed applications. Peter is also the author of Go kit, a toolkit for microservices... Read More →

Monday May 9, 2016 09:50 - 10:30 CEST

11:00 CEST

Of flags, experiments, and the power of etcd
At Improbable we’re building a platform for the next generation of Simulation, called SpatialOS. We’re hosting large scale Deployments of user Simulations inside our Clusters. Due to the real-time nature of Simulations, we needed a mechanism that would allow dynamic reconfiguration of parameters across multiple nodes of a Deployment. We leveraged etcd consistency and watch API to build a mechanism that allows developers to easily define code variables (Java annotated fields or Go flags) that change in almost real-time across hundreds of machines. While the main driver behind this highly flexible approach was our unique Simulation Runtime, the same mechanism is now powering our Platform’s operational configuration: request routing, rate limiting and experimental A-B testing.

avatar for Michal Witkowski

Michal Witkowski

Head of Infrastructure, Improbable
Michal is the Head of Infrastructure at Improbable. He and his team are building the platform of SpatialOS, a PaaS offering for strong simulation. The platform enables near real-time simulation of spatially distributed problems by facilitating cooperation of off-the-shelf workers... Read More →

Monday May 9, 2016 11:00 - 11:40 CEST

11:50 CEST

Elastic etcd – automatic add, replace and cluster growth
While etcd was born in the cloud era, it does not really play well in a dynamic environment where nodes come and go and where IP addresses are ephemeral. Moreover, etcd is meant – with its RAFT algorithm at the core – as a consistent key-value store. It rather refuses to form or join a cluster than putting consistency at risk. This talk will discuss how to make etcd more elastic in a sense that cluster join, member replacement and cluster growth is handled automatically when etcd starts up. The existing discovery service is used for this, but in contrast to the stock etcd not only during bootstrapping, but during the whole life-cycle of an etcd cluster. The ideas are implemented in the experimental elastic-etcd wrapper around etcd2. Its usage will be shown and how it behaves in certain cluster situations.

avatar for Stefan Schimanski

Stefan Schimanski

Senior Software Engineer, Red Hat
Stefan is a Principal Software Developer at Red Hat working on Kubernetes and OpenShift, with a focus on API machinery, extension points and developer tools as part of Sig API Machinery. He is one of the top-3 reviewers in the project (number of reviews). Stefan is a 2nd time GoogleSummer... Read More →

Monday May 9, 2016 11:50 - 12:30 CEST

13:30 CEST

dgr : a tool to build and run rkt container at scale
At BlaBlaCar we successfully run all our services with CoreOS and rkt. To achieve this, we had find a way to build efficiently our containers. Before that we had all our infrastructure managed by Chef and we didn't want to loose the power of config management along the way. This talk will be about the path we've taken, the mistakes we've made and why we ended up coding our own building tool : dgr.

avatar for Simon Lallemand

Simon Lallemand

Architect, Bl
Simon is a systems engineer at BlaBlaCar on the team in charge of the containers infrastructure. Simon is passionate about revolutionizing the way systems are managed to always go faster, be more reliable and scale.

Monday May 9, 2016 13:30 - 14:10 CEST

14:20 CEST

CoreOS and rkt app workflows with NodeJS
In this half hour, we'll explore the utility of manipulating and controlling images with NodeJS applications. Starting by deploying to a new Azure VM using a NodeJS built controller, we log in and study the rich development potential of NodeJS and NPM inside the new container. From there, we explore interfaces to rkt and finally consider backup and cluster scenarios built using NodeJS as well.

avatar for Michael Schloh von Bennewitz

Michael Schloh von Bennewitz

Michael Schloh von Bennewitz is a computer scientist specializing in network engineering, mobile design, and telecom server development. Responsible for research, development, and maintenance of packages in several community software repositories, he actively contributes to the Opensource... Read More →

Monday May 9, 2016 14:20 - 15:00 CEST
Tuesday, May 10

09:50 CEST

The Container Ecosystem and CoreOS
Tuesday May 10, 2016 09:50 - 10:30 CEST

11:00 CEST

Scaling fleet Clusters with gRPC
Fleet is a distributed init system that we use to schedule both infrastructure as well as user services on our clusters. Fleet is one of the main building blocks of our architecture that we use to schedule both infrastructure as well as user services on our microservices infrastructures. Fleet has seen production use for some time and is largely considered stable. The versatility and simplicity of fleet offers a distributed orchestration system, on which services in the form of systemd units (optionally running containers), can easily be deployed on the cluster. However, fleet can run into scalability issues that grow when using it in large-scale infrastructures. In particular, we realized of these scalability issues when fleet has to coordinate many nodes and/or thousands of deployed units in a cluster. Thus, we set out to improve the scalability of this distributed init system. Through the (optional) use of gRPC for communication between the fleet nodes as well as reducing etcd load through storing transient data in-memory, we could improve fleet performance and stability significantly. This talk goes into details of our design decisions and the implementation of the performance-improving changes, which we are currently working on merging upstream.

avatar for Hector Fernandez

Hector Fernandez

Engineer, Giant Swarm GmbH
Héctor Fernández is a DevOps at Giant Swarm, a German startup that offers simple Microservices Infrastructures to host containerized microservices. At Giant Swarm, Héctor focuses on monitoring, performance, and improving Giant Swarm's infrastructure. Over the last years, he worked... Read More →

Tuesday May 10, 2016 11:00 - 11:40 CEST

11:50 CEST

Dockyard - Container Registry And Volume Management For rkt
Rkt is the fastest growing container engine followed the Appc spec and tight integration with Kubernetes, but it still need container registry and volume management for running. Dockyard is a registry engine supported rkt pull & push which comply app container image discovery of Appc spec and a volume management with distributed file system. Dockyard has supported Docker Registry V1 & V2 at the same time, and has converted Docker image to Appc ACI. It’s important key to reuse the ecosystem of Docker. It has vulnerability analysis used Clair and content analysis. It has multiple backend drivers support storage container images into cloud object storage like Google, Amazon and others service, and has own build-in simple object storage service with three replications written by Golang and embed C.

avatar for Quanyi Ma

Quanyi Ma

Open Source Experts, Huawei
Quanyi Ma is open source expert in Huawei Corporation, senior architect and open source container & Golang evangelist. He interested in Container, DevOps, Golang and AngularJS. Ma also is a organizer and host of Docker meetup in many cities in China, member of some container conference... Read More →

Tuesday May 10, 2016 11:50 - 12:30 CEST

13:30 CEST

Next Generation Config Mgmt
A prototype for a next generation config management tool, and the specific problems this design solves. Three of the main design features of the tool include:

* Parallel execution
* Event driven mechanism
* Distributed architecture

This talk will demo a prototype I've built that implements these ideas. It is written in golang, and is completely free software. This talk will feature working code and a number of live demos! The tools makes heavy use of etcd, and the talk will focus on how we embed the project and the unique way in which we use it.

An introductory blog post on the subject is available. https://ttboj.wordpress.com/2016/01/18/next-generation-configuration-mgmt/ Attendees are encouraged to read it before the talk if they are interested!

avatar for James Shubin

James Shubin

Principal Software Engineer, Hacker
James is a DevOps/Config mgmt. hacker and physiologist from Montreal, Canada. He often goes by @purpleidea on the internet, and writes "The Technical Blog of James". He studied Physiology at university and sometimes likes to talk about cardiology.

Tuesday May 10, 2016 13:30 - 14:10 CEST

14:20 CEST

Namazu: a debugger for distributed systems specific bugs
Developing and maintaining distributed systems is difficult. We believe that the difficulties comes from lacking of a good debugger for the systems. Unlike non-distributed systems, bugs in the distributed systems are not understood well. Therefore building good debuggers is still difficult. The presenter will introduce Namazu project, an OSS framework of distributed systems fuzzing tester. Namazu focuses on two points of the systems: 1. error handling of hardware fault, and 2. non determinism of events e.g. packet arrival. They are distributed systems specific points and not covered by existing debugging techniques. The presenter will share success stories of removing bugs of real systems including etcd, Zookeeper, and Apache Yarn with Namazu.

avatar for Hitoshi Mitake

Hitoshi Mitake

NTT Labs
Hitoshi Mitake is a research engineer working for NTT Laboratories. He has been working on distributed storage systems and related technologies. These days, he is working on the etcd project actively. Current main topic is the design of authentication and access control of v3 API... Read More →

Tuesday May 10, 2016 14:20 - 15:00 CEST