Taking talks back to OpenStack: 20 Mirantis Presentations for Hong Kong Summit 2013
One of the things that's great about the semi-annual OpenStack summit is that it provides a different venue for bringing the community up to speed on work we've done across the broad variety of customers and OpenStack use cases across the open source cloud marketplace. Naturally, a lot of this works turns into code and commits, but there's still a lot to take and talk about on a broad range of subjects.
So for this fall in Hong Kong, our engineers and consultants have put together 20 different talks across a broad range of subjects we thought you'd find interesting, ranging from Neutron networking to Heat orchestration to budgeting to hypervisors and much more in between. If you agree they're interesting and you're a member of the foundation (join here to vote), we invite you to express your enthusiasm and help us bring these topics to the summit in Hong Kong.
- Vote Case Study: Running Windows on OpenStack – Georgy Okrokvertskhov, Das Kamhout | Summary
- Vote OpenStack Co-opetition: a view from within – Boris Renski | Summary
- Vote OpenStack Elimination Game – David Fishman, Kirill Ishanov, Randy Bias (CloudScaling), Mark Collier (OpenStack Foundation) | Summary
- Vote Savanna: Elastic Hadoop on OpenStack– Ilya Elterman, Sergey Lukjanov | Summary
- Vote Getting your code merged in OpenStack: the unwritten rules – Boris Pavlovic | Summary
- Vote Benchmarking OpenStack Performance at Scale – Boris Pavlovic | Summary
- Vote Does the Hypervisor matter in OpenStack? – Jason Venner | Summary
- Vote Step 1, launch OpenStack; Step Zero, pick your hardware – Greg Elkinbard | Summary
- Vote What if your OpenStack Deployment succeeds? – David Easter | Summary
- Vote Securing OpenStack for compliance – Tomasz Napierala | Summary
- Vote Cloud Vital Signs: Facilitating OpenStack operations – Szymon Banka, Kamil Kwiatkowski | Summary
- Vote OpenStack network virtualization review – Przemislaw Grygiel | Summary
- Vote Go Big, Go Fast: Scale-out SDN with OpenStack – Przemislaw Grygiel, Piotr Siwczak, Chloe Ma (Juniper) | Summary
- Vote Who is building OpenStack and Why – Project Stackalytics and many ways to slice the OpenStack Pie – Alex Freedland | Summary
- Vote What to do when the software is free but the cloud isn't – Anne Friend | Summary
- Vote Resource Reservation in OpenStack Cloud – Dina Belova | Summary
- Vote Educating Stakeholders on the value of SDN and OpenStack – Jason Venner | Summary
- Vote Parsing the current state of Neutron plugins – Damian Igbe | Summary
- Vote Neutron Network Namespaces and IPtables: technical deep dive – Damian Igbe | Summary
- Vote Load balancing in OpenStack – Ilya Shkhat, Eugene Nikanorov | Summary
Speakers: Georgy Okrokvertskhov, Das Kamhout (Intel)
With the rapid expansion of OpenStack cloud and growing enterprise footprint, it's not going to be long before it encounters another ecosystem with a big enterprise footprint: Windows. But there doesn't need to be a winner-take-all scenario. In fact, we see a growing demand to migrate and or run existing Windows based environments on an OpenStack cloud. In this talk we will cover a use case of using Windows Server on top of OpenStack. Our primary focus will be on automation for Windows in order to provide self-service features for OpenStack users. We want to share our experience with complex Windows environments automatic provisioning which is far beyond of simple Windows instance creation.
During this session we'll present the Intel IT use case and their view on the automation and self-service provisioning for Windows on OpenStack and the solution built by Mirantis for Windows automation on OpenStack.
Speaker: Boris Renski
My job at Mirantis is to figure out how to “co-opete” in this crowded space. Yes, we are all in OpenStack to build a great open source community. But let’s face it, we are REALLY in it to make money… So…. Does Rackspace regret letting go of control? Will Red Hat OpenStack mean death for the start-up distros? And what is the game plan for IBM and HP? I spend my days analyzing every move anybody makes in the ecosystem and figuring what that means for the industry. In this talk I’ll shed some light on OpenStack coopetition politics, share subjective view on strategies of individual players, and offer some predictions on future OpenStack competition landscape.
Host: David M. Fishman
Panelists: Mark Collier, Randy Bias, Kirill Ishanov
During this session we’ll productively use our time to give away 5 iPads to the audience in a trivia spectacular we call “The OpenStack Elimination Game.” We’ll present difficult OpenStack technical questions to a panel of 3 experts. One of them will provide the correct answer; the other two will bluff.
Everybody in the audience gets flashcards do hold over their head and vote which elimination sessions, each displaying a series of various questions related to OpenStack history, technology and community up on the screen. Each question will have 2-3 answer options. Raise your flashcards to identify which answer is right. Get it wrong – you are out. Last one standing gets an iPad ….and fame associated with being an OpenStack guru.
Speakers: Ilya Elterman, Sergey Lukjanov
This talk is about provisioning and managing Hadoop clusters on OpenStack using Savanna project. Savanna supports two key use cases: on-demand cluster provisioning and on-demand Hadoop tasks execution (Elastic Data Processing). This presentation will be focused around EDP functionality.
We’ll give an introduction to Savanna project, review features implemented on 0.3 version and talk about further plans, cover key architectural aspects and make the live demo. In this demo we’ll show how user can execute Hadoop job in one click on data stored in Swift using pre-configured Hadoop cluster template.
Speaker: Boris Pavlovic
Do you have a lot to contribute to OpenStack, but can't seem to get your code merged? Your commits are gathering dust because no one is reviewing them? Or maybe you have the resources to contribute, but don't know where to start? In this session, Boris Pavlovic of the Mirantis Contributions team, who helped drive Mirantis into the top 5 contributing organizations for the Havana release cycle, will cover:
- a step-by-step walkthrough of the process: Get started: where to commit and how to get your commits merged
- effective influence tactics and tricks for getting your work recognized in the community
- how to help people you've never met and get them to help you
- Don’t go there: mistakes you don’t want to make
- Keeping track of how others in the community are working
- Building and managing contributions teams
We’ll provide specific examples and step-by-step guidance.
Speaker: Boris Pavlovic
OpenStack at scale is like teenage sex; everybody talks about it, but nobody does it. Working with customers and partners, we’ve pushed OpenStack to its limits to see where it stumbles and where it breaks.
In this session, we’ll describe the details behind the process, talk about the tools performance assessment strategies we used. We'll also share our findings on key scalability and performance bottlenecks, validations approaches, and suggest solutions.
Speaker: Jason Venner
OpenStack developed largely with KVM; ESXi and VMware are widely considered one and the same. Is it possible to mix and match, and why would you want to? What are the opportunities and challenges of running OpenStack orchestration on an ESXi hypervisor? In this talk, we’ll look at some real world examples of hypervisor compatibility with OpenStack, examine some performance data and sensitivity analysis, and identify practical approaches to configuring an ESXi and mixed hypervisor environment in your OpenStack cloud.
The power of OpenStack software implementation, deployment and operations all cross the critical path of hardware selection. There’s a lot of apocryphal conventional wisdom on making hardware choices in the ecosystem and internet, such as “Q: How much memory do I need? A: More.” In this presentation, we’ll present a distillation of the key factors we have applied at Mirantis through dozens of highly successful OpenStack deployments on how to select an optimal hardware platform for your OpenStack cloud. Key factors we’ll cover include:
- Understanding users and audience
- Which requirements matter
- Making sense of costs
- Vendor Choices
- Looking ahead to the long Term Roadmap
The presentation features practical decision making and detailed technical considerations.
Speaker: David Easter
OpenStack efficiencies can be a two-edged sword. Sure, it can empower developers to invoke well-specified infrastructure services and build more cool stuff faster. But at the same time, the diversity of options and components in your infrastructure you want to deploy as a service means your cloud can get fragmented into messy underlying configurations and spin out of control: goodbye efficiency. And if your developers love your OpenStack cloud? Success is its own punishment.
How can you achieve reliable operation at scale in large distributed deployments, without sacrificing rapid adoption of new features? Much of the emerging technologies supporting OpenStack deployment and automation focuses on helping build the foundational elements of your OpenStack cloud. At Mirantis, across over 60 customer deployments, we’ve developed a set of pre- and post-flight checks to ensure that your OpenStack cloud can overcome that fragmentation. We’ll review those along with the best practices in deployment, monitoring, management, and maintenance. By taking advantage of this accumulated experience rather than re-inventing the wheel, you can speed the time to benefit for OpenStack. What’s more, once your cloud is up and running, these strategies mean that rather than firefighting, infrastructure teams can design/architect new services that developers and users use to capture value for their organization.
Speaker: Tomasz Napierala
OpenStack adoption is growing constantly, but there are still areas where vanilla OpenStack cannot be used out of the box. One of those fields is financial sector requiring infrastructure to be properly hardened and secured to meet challenging industry compliance requirements. To make OpenStack suitable for security demanding environments we’ve analysed most of the components and developed set of modules for system hardening. We’ll show a pragmatic set of guidelines to deploy OpenStack based cloud infrastructure able to meet most of PCI DSS requirements.
Speakers: Szymon Banka, Kamil Kwiatkowski
Success with OpenStack does not end with a successful deployment; operating your cloud requires a broad toolset that track the vital signs of your cloud. We’ll show our efforts derived from real customer engagement implementing OpenStack Operations portal using such technologies as Zabbix/StackTach/Graylog2. We will focus on showing how well-known can be leveraged to tackle common day-to-day pains of OpenStack administration, such as determining what user resources were affected by failures, tracking users actions on resources, capacity planning, alerting etc.
Speaker: Przemislaw Grygiel
Review of the Openstack common network virtualization technologies such as: VLAN, GRE, VxLAN, LISP, STT. During the review we will show how these protocols work we will also show what advantages we can get from particular protocols in various cases.We will focus on massive scale installations and challenges for network design. Where is the OpenFlow in the Openstack networking? How can we use it just for a network virtualization or it is something more?
Speakers: Przemislaw Grygiel, Piotr Siwczak; Chloe Jian Ma (Juniper)
OpenStack clouds are critically dependent on their network configuration, both with respect to ease of deployment, as well as flexibility of capacity allocation and assignment. Juniper Networks JunosV Contrail family of products delivers a complete network virtualization and intelligence solution that enables cloud automation and orchestration, and elastic service chaining of network and security services, while providing real-time view into network operations through a robust analytics engine. In this presentation, we’ll show how we use a collection of open-source tools and open standards to
- automate deployment of scale-out SDN in Openstack cloud
- seamlessly interoperate between physical and virtual environments
- abstract networking as a service
- enable dynamic inter-cloud federation and service chaining
[VOTE] Who is building OpenStack and Why – Project Stackalytics and many ways to slice the OpenStack Pie
Speaker: Alex Freedland
OpenStack has grown to prominence faster than any other open source project in history. As 71 companies with combined market cap of more than $1.2 trillion have contributed to OpenStack Havana release and over 700 developers committing code to Havana, the future of OpenStack look bright. But what are these players actually doing? Where they are committing code and why?
Project Stackalytics provides the ultimate transparency on what goes inside OpenStack. Literally! stackalytics.com shows the individual and company statistics as measured by the lines of code, number of commits, number of reviews and bug fixes. We will review some data samples and will discuss which statistics is more meaningful and will ask the community feedback as to what other contribution data it will make sense to measure. Last but not least, we will share some of fun facts from the statistics collected to-date.
Speaker: Anne Friend
OpenStack is free. Building and running a production cloud costs real money. Since going big can involve big spending, you'll need a plan – whether you use your OpenStack cloud for flexible capacity, for running ongoing business-critical applications, or a range of other possibilities.
In this presentation, I'll present a detailed plan for organizing your cloud budget spanning hardware, software, and wetware. In this presentation, we'll outline practical guidelines we've used at Mirantis to size cloud projects for a wide range of clients, so that you know what questions to ask in planning the cost side of your use case requirements. We'll also cover assumptions you can (and can't make), how to identify stakeholders and get their buy-in, and other practical tactics to use get the budget and business model right in advance.
Speaker: Dina Belova
OpenStack is growing in multiple directions, and new use cases emerge almost every day. For the use case of a large internal private cloud shared among wide range of users, we inevitably need to think about resource reservations – who gets to use what and when, and how to make sure that all of the sudden your cloud won't run out of resources:floating IP's, dedicated hosts with certain fancy peripherals, etc.
Moreover, if by chance you're running aggressively hungry loads and need to consider power consumption - you definitely want a service which will allow your users to present their plans on using certain amounts of resources, so that you could plan the underlying servers' availability/utilization accordingly.
It's for this reason we believe resource reservation could make OpenStack cloud more flexible and well managed. In this session we're going to talk about how resource reservation can work - what it means, why it's needed and how it will work in OpenStack based clouds in the foreseeable future.
We'll go through various types cloud resources that users may need to reserve and will discuss various opportunities, use cases and problems to be addressed: immediate reservations, future reservations, reclaiming of reserved resources and etc. At the end of this session we'll outline what has already been accomplished in this direction and what are the next steps to be taken.
Speakers: Jason Venner; Jon Vestal, PacNet
Even with the rapid increase of interest in OpenStack and Cloud technologies, SDN is still beyond the thought horizon of most organizations. How do you identify who benefits most from SDN, and educate potential stakeholders on the value of SDN implementation and design to their business? The challenge facing SDN pioneers is that traditional customers in the IT world tend to be highly risk averse, with operations driven by ticketing systems, human process controls and changes measured in calendar time, relying on a large staff to manage and execute the changes.
All the while, their end users are making end runs around the IT infrastructure, often violating compliance requirements and incurring uncontrolled spend with external providers. Ironically, while multi-DC SDN and Private or shared OpenStack Clusters are within reach of these organizations, the large conceptual gap both for traditional buyers and users of legacy IT infrastructure is difficult to cross. Even when they see in value of the OpenStack model and its operational efficiencies, the benefits of SDN can be elusive.
This talk is about how we close these conceptual gaps between the potential of new SDN technologies and existing patterns of purchasing and provisioning processes in enterprise networking. By understanding where key decision-making constituencies rightly or wrongly perceive risk, it's possible to significantly improve the likelihood that your SDN initiative – and its intended benefits – will succeed.
Speaker: Damian Igbe
OpenStack Neutron offers network-as-a-service for Openstack cloud infrastructure with a rich set of features for both the cloud tenants and providers. Just as nova compute provides IaaS with the ability to interface with a hypervisor of choice, Neutron also offers the ability to integrate the data center networking infrastructure with a neutron plugin of choice. Unlike the world of hypervisors, however, the world of neutron plugins can be confusing. Common questions include:
- Which neutron plugin should I use and why?
- What are the major differences between the plugins?
- Do all plugins use Open Virtual switch (OVS)?
- Should I choose a plugin that implements network overlays or Software Defined Networking (SDN)?
We’ll start with a technical deep dive into Neutron plugin architecture and then survey the current available plugins, expose the unique features, compare the similarities and differences especially in terms of the major design decisions. After this talk , attendees should have a clear idea of which neutron plugin might be most suitable for their Openstack cloud infrastructure.
Speaker: 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’s 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. We’ll show how to go about troubleshooting L3 issues, and how to apply this more robust networking abstraction in distributed OpenStack environments.
Speakers: Ilya Shakhat and Eugene Nikanorov
Put all bricks together for a production-ready load balancing: multi-vendor support in Neutron LBaaS, elasticity via Heat templates and service monitoring via Ceilometer. The session will cover new features introduced in Havana release cycle as well as briefly cover the way of writing vendor drivers for the service. The talk also will cover the roadmap planned for Icehouse.
Key highlights: data model change, LBaaS API extensibility, hardware appliance support, integration of LBaaS with Heat and Ceilometer.