Global OpenStack: the Power of Cloud Interoperability

Monty Taylor

http://inaugust.com/talks/global-interop.html

twitter: @e_monty

Who am I?

Office of Technology

Zuul

Ansible

Who am I?

Technical Committee

Developer Infrastructure Core Team

Former Foundation Board of Directors

PTL of shade project

I want to deploy and run an application on the internet so that my customers all over the world can consume it.

I want to deploy the application across multiple clouds or regions so that my service survives issues in any one of them.

I want to deploy the application in a location of my choosing so that I can comply with regulatory demands.

I want to write some new Cloud Native applications while continuing to make use of my existing investments.

I need to move faster than before AND I need increased stability at the same time.

THIS WORKS

I'm doing it myself as we speak

20k VMs per-day

20 Cloud Regions in 9 clouds

Only using OpenStack APIs

OpenStack Infra

Tooling and Automation for the development of OpenStack Project

>2500 Developers

Gated Changes

Every change goes through enforced automatic full integration testing (at least twice) before landing

Integration Testing

Install and run one or more clouds then validate that those clouds work.

Each Test Job Runs on Single Use Cloud Servers

This is that "cloud scale out" part

2 KJPH (kilo-jobs per hour)

Our Cloud Servers are everywhere

Public Clouds

  • Rackspace: Dallas, Chicago, DC
  • Internap: New Jersey, Montreal
  • OVH: Strassbourg, Gravelines
  • Citycloud: London, Los Angeles, Stockholm
  • Vexxhost: Montreal

Managed Private Clouds

  • Red Hat TripleO: Phoenix

Private Cloud

Infra Cloud

  • Hardware donated by HPE: Houston
  • Community run cloud using OpenStack Puppet
  • Bare Metal managed with Bifrost/Ironic

image

Gerrit

  • Code Review and Code Hosting
  • Traditional 'Enterprise' Java Application
  • Single OpenStack Nova VM, Cinder Volume
  • Scale out farm of git replicas

nodepool

  • Cloud Native resource manager
  • Purpose built in Python
  • Keeps a pool of ready to go nodes
  • Multi-cloud
  • Fully elastic - responds to demand

zuul

  • "Test it like you deploy it"
  • The Gatekeeper
  • Microservices but with a centralized scheduler
  • Responds to code review events from Gerrit
  • Runs Ansible content on nodes from nodepool
  • Use production Ansible to test proposed changes
  • Talk about Zuul Tomorrow 15:50-16:15 on stage OP5

OpenStack Works!

Why should you care?

Is interoperability really important?

Run what you want

Cloud-native application

Traditional Java web application

High-performance Database on Baremetal

Run it where you want

  • Put it near your users
  • Avoid putting it in an US data center
  • Locality Requirements
  • What if you have customers in China?
  • What if you also have customers in Sweden? New Zealand? Brazil?

Run it how you want

  • Public
  • Hosted Private
  • Managed Private On-Premise
  • Run your own with a partner/vendor
  • Run your own on your own

China’s Sinorail Information Cloud

Billions of Passengers

Anybody REALLY think that should just run in a Public Cloud?

Anybody REALLY think that should run in a Public Cloud owned and operated by a US company?

What about the China National Offshore Oil Corp?

or the State Grid of China?

OpenStack in Asia

Huawei: First Chinese Platinum Member of OpenStack Foundation

Only 8 of the 24 Gold Member Companies are non-Asian

European OpenStack Public Clouds

CityCloud: Sweden, UK, US

Datacentred: UK

Elastx: Sweden

Enter Cloud Suite: Italy, Germany

Fugo: Netherlands

Internap: Netherlands, US, Singapore

OTC: Germany

OVH: France, US

Switch: Switzerland

Ultimum: Czech Republic

Zetta: Norway

kiss.cloud: Netherlands

Other Public OpenStack Clouds

That I personally have an account on

Auro: Vancouver

Catalyst: New Zealand

Conoha: Japan, Singapore, US

Dreamhost: US

Ormuco: Canada, Federated Worldwide

Vexxhost: Canada

UnitedStack: China

The OpenStack Community is Designed for Collaboration

The OpenStack Project Believes all Participants are Equal

OpenStack isn't going to remove something you need...

because OpenStack IS you.

I'm in this to make the world better.

The whole world, not just my corner.

Why does Interop Matter?

Is Sinorail going to run China's Train system with an app they download from an App Store?

Nope

Are they going to let me run VMs for OpenStack Infra Nodepool on their Cloud?

Nope

That would be SUPER cool though!

Who cares if China National Offshore Oil Corp, State Grid of China and Deutsche Telecom's clouds work the same?

It's all about the tools

  • Ansible
  • OpenShift
  • Terraform
  • Spinnaker
  • Zuul

Ansible OpenStack Modules

  • Based on the API consumption code from nodepool: "shade"
  • Support all of the known OpenStack Public Clouds
  • Interoperable == you can use them with your cloud
  • Maintained by me
Talk about Ansible and OpenStack Today 16:00 - 17:00

Spending Resources Wisely

If China National Offshore Oil Corp, State Grid of China and Sinorail all run Interoperable OpenStack Clouds ...

they share not just the OpenStack Cloud software itself...

they ALSO share consumption and orchestration software...

and can focus MOST on Oil, Electricity and Trains


Oil, Electricity and Trains are important

With Interoperable Clouds

It's like I work for all of you!

It's like we all work with each other

China understands the power of Open Collaboration on shared problems.

Red Hat and Huawei are working together to bring you that power.

  • Delivering Red Hat Software on Huawei Clouds
  • Delivering Huawei VNFs on Red Hat Clouds

Retooling the Telco Network Infrastructure

NFVI Minimizes Harware Dependencies

for Network Simplicity and Flexibility

Open Source is the Foundation of the Modern Telco Network

OpenStack, Red Hat and Huawei give you the power to make the choices that are right for you, your business, your users, your country and the world.

Thank you!

http://inaugust.com/talks/global-interop.html

twitter: @e_monty