Register for the OpenStack Summit    Register Now

OpenStack Conference Fall 2013

November 5-8, 2013, Hong Kong

November 5, 6, 7, 8


The OpenStack Summit is a four-day conference for developers, users, and administrators of OpenStack Cloud Software.


The OpenStack Summit
Hong Kong 2013

Videos of Sessions From Day 4

OpenStack Networking Hands-on Lab

Eric Lopez,Aaron Rosen

Users will get access to a live OpenStack + Neutron setup and be able to walk through key neutron deployment use cases, with members of the Neutron core development team available to provide guidence and answer questions. At the past two OpenStack conferences we presented a similar Neutron hands on lab led by several members of the Neutron core team and it was standing room only. We'd like to run another session this time, incorporating lessons learned from the previous sesion and also including new Neutron capabilities introduced in the Havana release. Demonstrated features will include: private L2 networks using tunnels rather than vlans. Including support for overlapping IPs. L3 + NAT via neutron logical routers Firewall as a Service VPN as a Service Loadbalancer as a Service and more!

Watch Now

Putting the PaaS in OpenStack with Heat

Diane Mueller,Krishna Raman,Chris Alfonso

Just getting a cloud environment up and running is no longer enough. The challenge that OpenStack faces is how to get people, applications and services working on OpenStack out of the box and to ensure that the “unboxing” experience is as seamless and painless as possible. Organizations expectations for deploying cloud now include being able to rapidly make services and applications available as soon as they have IaaS deployed. To meet the expectations of most organizations, adding a PaaS layer has become an essential part of every Cloud deployment strategy. The OpenShift Origin PaaS project is backed by the fastest growing open-source community of developers, cloud architects, devops, and end users intent on creating the next generation of PaaS and ensuring that the tools for deploying, managing & scaling it for OpenStack are freely available. To do this on OpenStack, the OpenShift community has “adopted” Heat, OpenStack's orchestration engine, and delivered a set of Heat templates for deploying, managing and auto-scaling OpenShift on any OpenStack distribution. This talk provides an overview of OpenShift, RedHat's Platform as a Service and a deep dive into deploying OpenShift using Heat, OpenStack's template language, and a live demonstration of Heat technologies to deploy AND autoscale OpenShift using repeatable orchestration templates. We will demonstrate the power of Heat, OpenStack's orchestration engine and how we leverage Heat to orchestrate cloud infrastructure resources such as storage, networking, and instances to deploy OpenShift into a repeatable running environment for OpenStack IaaS platforms. OpenStack Summit attendees can learn about both the OpenShift Origin Project, and the emerging Heat template technologies and its impact on Linux and open source cloud communities. The speakers are both experienced with live demonstrations, and make the technical difficulty of this topic easily approachable through real-life examples. Speakers: Diane Mueller, Krishna Raman & Chris Alfonso

Watch Now

Highly Available MySQL for OpenStack

Jacob Walcik,Rags

The primary requirements for OpenStack based clouds (public, private or hybrid) is that they must be massively scalable, highly available and so on. MySQL is the basis for many of these OpenStack deployments since critical components like Keystone and Nova depend on it. Attend this session for a technical overview of how to make MySQL highly available, using either the a) MySQL/Galera or b) DBRD/Pacemaker and c) MySQL cluster with HAProxy, keepalived and VRRP approaches. The presentation compares and contrasts the approaches and discusses some best practices and recommendations. After attending this session, attendees will have a good perspective on making MySQL highly available for their OpenStack implemenations.

Watch Now

OpenStack Deployment with Chef Workshop

Justin Shepherd,Matt Ray

Want to learn more about using Chef to deploy OpenStack and manage infrastructure on top of it, but not sure where to start? This in-depth, hands-on deployment session will cover the Chef and OpenStack ecosystem and how to get started with the Chef cookbooks in the StackForge repositories. We'll cover the current Grizzly OpenStack resources and the related cookbooks and content in the Chef community. Topics covered will include: * Deployment configuration and techniques * StackForge repository code walkthrough * Cookbook development and testing * Deploying and managing infrastructure on OpenStack with the knife-openstack plugin * Documentation The session is intended for folks already familiar with Chef and interested in deploying OpenStack. This is intended to be a very interactive session with many questions and guided code and deployment walkthroughs. Attendees are expected to provide their own laptops capable of running a single-node OpenStack virtual machine. Want to learn more about using Chef to deploy OpenStack and manage infrastructure on top of it, but not sure where to start? This in-depth, hands-on deployment session will cover the Chef and OpenStack ecosystem and how to get started with the Chef cookbooks in the StackForge repositories. We'll cover the current Grizzly OpenStack resources and the related cookbooks and content in the Chef community. Topics covered will include: Deployment configuration and techniques StackForge repository code walkthrough Cookbook development and testing Deploying and managing infrastructure on OpenStack with the knife-openstack plugin Documentation The session is intended for folks already familiar with Chef and interested in deploying OpenStack. This is intended to be a very interactive session with many questions and guided code and deployment walkthroughs. Attendees are expected to provide their own laptops capable of running a single-node OpenStack virtual machine.

Watch Now  Slides

Ceph: The De Facto Storage Backend for OpenStack

Sebastien Han

For more than a year, Ceph has become increasingly popular and saw several deployments inside and outside OpenStack. The community and Ceph itself has greatly matured. Ceph is a fully open source distributed object store, network block device, and file system designed for reliability, performance, and scalability from terabytes to exabytes. Ceph utilizes a novel placement algorithm (CRUSH), active storage nodes, and peer-to-peer gossip protocols to avoid the scalability and reliability problems associated with centralized controllers and lookup tables. Since Grizzly, the Ceph integration gained some good additions: Havana definitely brought tons of awesome features. It also made the integration easier and definitely removed all the tiny hacks. All these things, will certainly encourage people to use Ceph in OpenStack. Ceph is excellent to back OpenStack platforms, no matter how big and complex the platform.. The main goal of the talk is to convince those of you who aren't already using Ceph as a storage backend for OpenStack to do so. I consider the Ceph technology to be the de facto storage backend for OpenStack for a lot of good reasons that I'll expose during the talk. In this session, Sebastien Han from eNovance will go through several subjects such as (depends on the time that I'll get): Quick Ceph overview (for those of you who are not familiar with it) Quick state of the integration with OpenStack (general state and Havana's best additions) Quick cinder drivers overview and comparisons Building a Ceph cluster - general considerations Use cases and design examples (hardware harware hardware) Achieve HA with Ceph Operations: backups, monitoring, upgrades Tips and best practices

Watch Now  Slides

LDAP Integration for Private Clouds

Ryan Lane

LDAP integration in OpenStack can go way past just basic integration with Keystone. This presentation will discuss LDAP integration for private clouds for global users and groups, per-tenant local users and groups, per-tenant sudo, per-tenant autofs, and alternative methods for storing and using SSH keys. I'll also cover how to extend multi-tenancy information into other applications like Gerrit, puppet, and saltstack using LDAP. The majority of these examples use a common pattern which can be extended to most applications, I'll discuss the pattern and how you can use it in your own non-OpenStack applications.

Watch Now

Panel: Ideas, Tools and Examples for OpenStack Groups

Kyle Mestery,Nati Shalom,Colin McNamara,Sean Roberts,Shannon McFarland

Speakers: Sean Roberts, Colin McNamara, Kyle Mestery, Shannon McFarland, Nati Shalom Moderator: Tom Fifield

Watch Now

Getting Your Blueprint Accepted Quicker: the VPNaaS Use Case

Loic Dachary,Swaminathan Vasudevan

VPNaaS was a complex blueprint and back in April 2013, three or more blueprints were submitted on the same topic. It was difficult to come to a consensus since each submitter had different views and priorities. Despite of these difficulties, VPNaaS was agreed by all parties and scheduled for Havana a few weeks later. Upstream University is a training program designed to help Free Software contributors get their feature and patches accepted quicker. It played a small but essential role, behind the scene, in the acceptance of the VPNaaS blueprint. The story will be told, with testimonials from the people involved and a few anecdotes.

Watch Now  Slides

OpenStack Neutron Modular Layer 2 Plugin Deep Dive

Kyle Mestery,Robert Kukura

This presentation introduces the Havana release's new Modular Layer 2 (ML2) plugin for OpenStack Neutron. The ML2 plugin is a community-driven framework allowing OpenStack Neutron to simultaneously utilize the variety of layer 2 networking technologies found in complex, real-world data centers. ML2 currently works with the Open vSwitch, Linux Bridge, and Hyper-V L2 agents, and is intended to replace and deprecate those agents' monolithic plugins. The ML2 plugin also works with SDN controllers and network hardware devices, and is designed to greatly simplify adding support for new L2 networking technologies into OpenStack Neutron. In this session, Cisco and Red Hat representatives will: Introduce the Modular Layer 2 (ML2) plugin for OpenStack Neutron Provide an overview of ML2, discussing its design principles and detailing use case examples Describe ML2's architecture and its driver APIs Demonstrate an OpenStack deployment with ML2 utilizing multiple segmentation methods and multiple L2 networking mechanisms to show the power of the ML2 plugin Attendees will leave this session with an understanding of ML2, the use cases it was designed to solve, how to deploy ML2 in an OpenStack Havana environment, and how existing Neutron deployments can migrate to ML2.

Watch Now  Slides

Hands-on with OpenStack + vSphere

James Page,Gary Kotton

Nova vSphere support was added in Grizzly and enhanced in Havana. Likewise, Havana includes new support for a Cinder driver that uses vSphere datastores. Come to this hands-on workshop to learn more about how VMware vSphere works with OpenStack! In this session, each small group of 2-3 people will get access to a remot lab environment that consists of: - An OpenStack “controller” node. - A windows host running vCenter - Several ESX hypervisors. - A host provided shared storage. The session will walk you through the key steps in configuring the system for use with vCenter, provisioning servers + volumes using standard OpenStack interfaces, and viewing the resulting changes via vCenter to understand how the Nova + Cinder drivers for vSphere consume capacity from the underlying vCenter-managed infrastructure. We will also highlight some troubleshooting capabilities enabled by the use of an OpenStack-aware plugin for vCenter. An Internet connected laptop with a standard browser is required for this session.

Watch Now

Real-Time Telemetry Processing: Metrics, Logging and Dashboards

Brian Cline

While OpenStack projects provide a variety of ways to be notified of or extract telemetry-based metrics and usage information, today there exists no singular way of capturing both performance metrics and congruently logged information that correspond along the same time scale. In many cases it’s extremely valuable to have performance metrics and congruent log patterns collected, analyzed, and paired together in real-time.This talk will explore the use of Riemann (, a distributed systems monitor capable of handling tens of thousands of events per second, per core. Its abilities extend beyond simple metrics into service, state, tagging, descriptions (mayhap log messages), and can also scale beyond a single node; it is possible to construct a topology of Riemann servers that filter and pass events to each other for processing.A separate web-based dashboard project has thus far been capable of receiving thousands of updates per second via websockets, and is a good demonstration of Riemann's capabilities.Discussion and demo will illustrate how we can publish system-level metrics and Ceilometer-collected data to Riemann, while also collecting OpenStack log output--all while acting upon the following:*specific log patterns*service states*inactivity detection*activity rates for types of events*metric rates, values, and percentilesThis data can be normalized, analyzed, graphed, used to trigger automated first-response/front-line actions, sent to other systems, retrieved through a simple API to be presented within other systems, presented in other dashboards, and so forth.No previous experience with Riemann or Ceilometer is necessary for this session.

Watch Now  Slides

Towards Truly Open and Commoditized Software Defined Networks In OpenStack

Jun Park

EIG/Bluehost has been successfully managing one of the largest Openstack environments with more than 17,000 compute nodes wherein over 20,000 instances are running for a year now. We were happy to share some of our experiences and findings at the Portland summit and are grateful to see that many of our concerns have been aggressively addressed in the Havana release. However, we think that there is still a significant lack of SDN functionality which is open source, free, and does not require specialized networking equipment. More specifically, one of the default Neutron plugins, OpenvSwitch, has not changed much since the Folsom release. In this talk, we would like to share how we have been developing our own SDN plugin for our production environment at Bluehost. We hope to share our experiences and designs with the community so that we can facilitate the discussion towards truly open and commoditized SDN for the masses.

Watch Now  Slides

State of the Stack

Randy Bias,Scott Sanchez

OpenStack in three short years has become one of the most successful,most talked about and most community-driven Open Source projects inhistory.In this joint presentation Randy Bias (Cloudscaling) and Scott Sanchez(Rackspace) will examine the progress from Grizzly to Havana and delveinto new areas like refstack, tripleO, baremetal/Ironic, the move from"projects" to "programs", and AWS compatibility.They will show updated statistics on project momentum and a deep diveon OpenStack Orchestrate (Heat), which has the opportunity to changethe game for OpenStack in the greater private cloud game. The duo willalso highlight the challenges ahead of the project and what should bedone to avoid failure.Joint presenters: Scott Sanchez, Randy Bias

Watch Now

Database Clusters as a Service in OpenStack: Integrated, Scalable, Highly Available and Secure

Ivan Zoratti

MySQL, MariaDB or Percona Server are the de-facto standard for OpenStack in terms of internal database and DBaaS. Today, MySQL is offered as a single instance, typically unintegrated in the IaaS layer, or in the best case it is provided with standard replication, but without full control on availability and failover. Last but not least, MySQL security is not combined with the security mechanisms in the cloud.This presentation is an introduction to MariaDB for OpenStack, with an extended set of API that automates the provisioning, deployment and configuration of a all-active set of servers with Galera synchronous replication.We will cover these topics:    • MySQL, MariaDB and OpenStack - what is the current status    • Deploying a MariaDB cluster in OpenStack: Glance & Nova integration, Red Dwarf compatibility and issues    • Automatic provisioning using Juju, Puppet and Chef    • DBA Daily operations and integration with Object and Block storage    • High Availability components: replication and automatic failover, HA options    • Security of the cluster: Keystone integration, database secure connections and tunnelling

Watch Now

OpenStack User Personas: 6 Months On

Dave Neary

Earlier this year, Dave Neary presented the theory of personas to attendees of the OpenStack Summit in Portland. Attendees were excited about creating a set of personas for the OpenStack project, as they allow you to have a much clearer idea of your target audience, what their needs are, and how you can reach them. They also allow much easier communication around feature discussions, user interface design and marketing strategy. Based on data from the user committee survey and user interviews, a personas working group is being created to answer the question: “Who uses OpenStack?” In this session Dave will return to present an initial set of OpenStack personas, discuss how they were created, and detail what conclusions we can draw from them.

Watch Now  Slides

Sheepdog: Yet Another All-In-One Storage for Openstack

Liu Yuan

Sheepdog is purely userspace distributed storage system for QEMU. It is essentially an object storage system that manages disks and aggregates the space and performance of disks linearly in hyper scale on commodity hardware in a smart way. On top of its object store, sheepdog provides elastic volume service (support of glance and cinder has been merged) and http service (in the development, plans to be Swift API compatible). Sheepdog doesn't assume anything about kernel version and can work nicely with a xattr-supported file systems. In this presentation, I'll concentrate on the technical aspects of sheepdog: 1. How sheepdog works internally regards of thin-provisioning volume, snapshot, clone and node managements. 2. What sheepdog can provides for Openstack. 3. Some performance numbers. 4. Demo of live sheepdog cluster

Watch Now  Slides

OpenStack Queuing and Notification Service (Marconi)

Allan Metts,Kurt Griffiths,Flavio Percoco

Come to this session to get an update on Marconi, an OpenStack queuing and notification service described at Marconi aims to be pragmatic, building upon the real-world experiences of teams who have solid track records running and supporting web-scale message queuing systems. Users can customize Marconi to achieve a wide range of performance, durability, availability, and efficiency goals. As a message bus, Marconi allows cloud developers to use a REST API to easily distribute tasks to multiple workers across the components of an OpenStack deployment. Publish-subscribe semantics are also supported, allowing notifications to be distributed to multiple listeners at once. Join Rackspace's Kurt Griffiths, Principal Architect, and Allan Metts, Engineering Director, to learn about the work that has been done and the path ahead -- including a description of the project, real-world performance metrics, and a live demo.

Watch Now  Slides


Nick (Nicolas) Barcet,Eoghan Glynn,Julien Danjou

Ceilometer is a tool that collects usage and performance data, while Heat orchestrates complex deployments on top of OpenStack. Heat aims to autoscale its deployments, scaling up when they're running hot and scaling back when idle. Ceilometer can access decisive data and trigger the appropriate actions in Heat. The result of these two OpenStack projects meeting is value creation in the form of an alarming API in Ceilometer and its consumption in Heat. In this session, we will detail how the two projects work together to deliver autoscaling, providing both background information and a technical deep dive.

Watch Now  Slides

Panel: International Forum: The State of OpenStack Around the World

Sajid Akhtar,Trung The Nguyen,Tristan Goode

Speakers: Tristan Goode, Trung The Nguyen, Sajid Akhtar, Moderator: Loic Dachary

Watch Now  Slides

The Elephant in the Room Hypervisor Breakouts

Rob Clark

Often ignored or hidden away in risk registers the consequences of hypervisor breakouts are incredibly high. In this presentation I describe potential exploitation vectors in common virtualization stacks before diving into hands on, practical guidance for securing your hypervisor and addressing breakout vulnerabilities when they occur.

Watch Now

Heat and Enterprise Application

Andrew Trossman,Thomas Spatzier,Jeff Sloyer,Jun Jie Nan,Edmund Troche,Lakshminarayanan Renganarayana

Deployment of large enterprise applications is a complex problem. Such applications consist of numerous software components spread across multiple VMs with a variety of dependencies between them, and require a large number of configuration parameters to be specified. Pattern technologies such as OpenStack Heat help automate the deployment of distributed applications. However, Heat in its current form only has limited supported for software orchestration to address the aforemenionted issues. Furthermore, especially for application components initial deployment is just the begining and more orchestration is necessary throughout the comlete lifecycle of an application to consider aspects such as high availability, storage and network configurations. In this session we will discuss automated deployment and maintenance of enterprise applications using OpenStack Heat. Our talk will foucs on the following issues and how they can be addressed using base Heat capabilities together with extensions to the Heat engine. 1. Enterprise application deployment: 1.1. complete declarative specification of application model; 1.2. cross component dependencies; and 1.3. software stack configuration and coordination.2. Enterprise application maintainance: 2.1. virtual machine high availability; 2.2. storage; and 2.3. network.

Watch Now  Slides

Neutron Network Namespaces and IPtables: Technical Deep Dive

Damian Igbe

Networking with OpenStack Neutron requires a different mindset around IP networking than conventional physical topologies. Understanding Linux namespaces is critical to troubleshooting Openstack Neutron networking and understanding of neutron network topology. A key element of this shift is use of Linux network namespaces, introduced in Folsom. What more, without a thorough understanding of how namespaces organize and abstract L3 routers, DHCP servers within the network and subnets spaces, network-induced downtime can be difficult to resolve. Namespaces enables multiple instances of a routing table to co-exist within the same Linux box (like virtual routing and forwarding (vrf) in routers), within the network and subnets spaces, per tenant. It introduces a whole realm of networking flexibility, which can be critical in production Openstack deployments -- but can also contradicts the logic applied by experienced IP network admins and lead troubleshooting off a cliff. This technical deep dive into Openstack Neutron Namespaces and IPtables wil give attendees will get a clear understanding of these building blocks of OpenStack L3 and DHCP agents. Wel show how to go about troubleshooting L3 issues, and how to apply this more robust networking abstraction in distributed OpenStack environments.

Watch Now  Slides

Making Your Apps Cloud Aware

Sebastian Stadil

Cloud Management used to be about launching servers and monitoring them, potentially autoscaling them. This was all good, but it not nearly enough! We need to make applications Cloud Aware! The idea is that your applications need be ready to be deployed in radically different contexts, and they need to understand the context theye running in. In this talk for application infrastructure developers (aka devops), we explain what is required to adapt traditional (or hastily built) cloud application infrastructure into Cloud Aware applications, so you fully leverage the power of OpenStack.

Watch Now

Panel: Measuring OpenStack: What Activities We Track and Why

Dan Stangel,Daniel Izquierdo,Alex Freedland,Qingye Jiang

Speakers: Daniel Izquierdo, Alex Freedland, Dan Stangel, Qingye Jiang Moderator: Stefano Maffulli

Watch Now  Slides

StackTach - Ceilometer Integration

Sandy Walsh

Over the last six months Rackspace has made major improvements to the StackTach monitoring application. Much of this has focused on reconciliation of notifications with backend billing systems and extensive reports for error detection. If you are a service provider you'll be interested in learning more about StackTach. This session will provide an overview and demo of the StackTach application and these recent improvements. Also, we'll discuss our strategy and progress on bringing this functionality into Ceilometer.

Watch Now  Slides

Is Open Source Good Enough? A Deep Study of Swift and Ceph Performance

Jiangang Duan

With the rising of cloud computing and big data, more and more companies start to build their own cloud storage system. Swift and Ceph are two popular open source software stacks, which are well deployed in today's OpenStack based cloud environment to implement object and virtual block service. In this presentation, we do a deep study on both swift and Ceph performance on Commodity x86 platform, including testing environment, methodolgy and thorough analysis. Tuning guide and optimization BKM (best known method) will also be shared for reference.

Watch Now  Slides

Bringing together Linux-based Switches, VLANs, and Neutron

Chet Burgess,Nolan Leake

Many OpenStack deployments use VLANs to separate traffic into different virtual networks. Existing solutions require either that all VLANs must be trunked to every host or the use of vendor-supplied plugins paired with specific (proprietary) hardware.Cumulus Networks and Metacloud will be presenting a plugin (targeted for inclusion in Icehouse) that allows Neutron to extend the configuration of VLANs to Linux based switches such that individual VLANs are only trunked to ports and hosts that need them. The use of Linux based switches allows for rapid prototyping of advanced configurations in VMs or on real hardware. We will be demonstrating this plugin controlling bare-metal hardware-accelerated Linux switches running Cumulus Linux for production ready deployments. Additionally, we will discuss an L3 mode, that has similar functionality but uses entirely L3 concepts. The discussion will include solutions to issues around connectivity, security, mobility, and public IP access.

Watch Now  Slides

Developing a Roadmap for Global Adoption: Exploring OpenStack's Globalization Best Practices

Tom Fifield,Ying Chun (Daisy) Guo

As OpenStack community grows, globalization becomes be an important point to the success of the community. There have been some discussions and practises in the community, in order to promote the global adoption of OpenStack.This presentation will summarize the existing activities in code development, documentation, translation process, and even community building in OpenStack community, the studies of best practises in other communities (e.g. GNOME, CloudFoundary), and then explore a roadmap as future plan.

Watch Now  Slides