Pairs OpenStack Summit Summary
- 1. Paris OpenStack Summit Recap Guang Ya Liu liugya@cn.ibm.com Sina Weibo: @platformer
- 2. Open Source is the Infrastructure of the Cloud Tomorrow
- 3. Overview • Over 5000 attendees, a significant number of enterprise attendees. • Over 185 Sessions, over 184 design summit sessions, and over 26 Ops summit sessions. • 45 major sponsors (large companies) and another 43 startup sponsors. • OpenStack is now: 17,300 People, 433 supporting companies, contributions from 145 countries, in excess of 20M lines of code • More than 4,600 developers & users in attendance – 33% - Product Strategy/Management / Architect – 26% - Developer – 16% - User / Operations SysAdmin – 9% - CEO / CIO / IT Manager – 10% - Business Development / Marketing – 6% - Other • Percent of 1st timers – 58% • Countries represented – 62 • Companies represented – 876
- 4. Major themes for Paris • 11 core projects have been defined in the 10th release of OpenStack (Juno). • The community is growing significantly with major companies joining – Intel is now a platinum member, and SAP has joined. • Docker and OpenStack is coming to one ecosystem • There is increasing interest around NFV within OpenStack - across carriers and service providers • Ceph is leading the unified storage solution for OpenStack
- 5. Docker All the OpenStack Service • Isolated, lightweight, and portable • Pre-integrated • Easily describe run-time relationships • Run on something thin and easy to update http://docker.com/
- 6. Docker All the OpenStack Service • Isolated, lightweight, and portable • Pre-integrated • Easily describe run-time relationships • Run on something thin and easy to update http://docker.com/ http://kubernetes.io/
- 7. Docker All the OpenStack Service • Isolated, lightweight, and portable • Pre-integrated • Easily describe run-time relationships • Run on something thin and easy to update http://www.projectatomic.io/ http://docker.com/ http://kubernetes.io/
- 8. Demo • OpenStack deploy Kubernetes cluster • The Kolla project is part of the OpenStack TripleO effort, focused on deploying OpenStack environments using Kubernetes and Docker containers. • Kolla call kubernetes to create another openstack cluster by running all services in docke containers • All applications can run in docker container
- 9. Orchestrating Docker with OpenStack http://allthingsopen.com/
- 10. Solum • An OpenStack related project designed to make cloud services easier to consume and integrate into your application development process. • Solum is natively designed for OpenStack clouds and leverages numerous OpenStack projects, including Heat, Keystone, Nova, Trove, and more.
- 11. HEAT Docker Resource • Heat provides a Docker resource • Docker resource communicates directly to Docker Server • Templates may glue Nova and Docker resources • Can deploy containers on top of VMs or bare-metal instances.
- 12. Hybrid Nova Configuration + Ironic
- 13. Future - MAGNUM • Container As A Service – Might be handled in a new incubation project – Gantt help resource scheduling – Create instance on demand – Container Agent – Agent Reply
- 14. Docker Best Practise • I have a lot of data – Do not put “data” in containers – Replicate your data in each target cloud – Bind mount your data volume(s) to your container(s) • I have a lot of cloud servers – You need a utility with a resource scheduler • Nova, Magnum, Kubernetes, Mesos etc – Treat application servers like cattle • Create them from container images using automation
- 15. Docker With Multiple Clouds • Docker Hosts Create – Automatically set up new docker hosts – Select which docker hosts to interact with – Support Azure, Digital Ocean and Rackspace – http://github.com/docker/docker/issues/8681 • Docker Clustering – http://github.com/docker/docker/pull/8859 – Support for running Docker on multiple hosts, with autoscaling capabilities and pluggable scheduling algorithms. – Similar with Kubernetes
- 16. Clocker • Clocker creates and manages a Docker cloud infrastructure. • Clocker support single-click deployment and runtime management of multi-node applications that can run on containers distributed across multiple hosts • Using the Weave SDN to manage network • Application blueprints written for Brooklyn can be deployed to a distributed Docker Cloud Infrastructure.
- 17. Clocker • Clocker – Brooklyn + Docker + Jclouds + Weave • Solves – Docker Networking – Container Placement – Application Definition
- 18. Docker Ecosystem • PaaS based on Docker • Kubernetes + Mesos – Mesos help Kubernetes do resource scheduling – https://github.com/me sosphere/kubernetes- mesos
- 19. RedHat: ManageIQ->CloudForm • Advanced virtualization management – Monitoring and tracking – Capacity management and planning – Resource usage and optimization – Workload life-cycle management – Policies to govern access and usage • Private or hybrid cloud management – Self-service portal and catalog – Controls to manage request – Quota enforcement and usage – Chargeback and cost allocation – Automated provisioning • Operational visibility and control – Dashboards – Reports – Policies – Alerts – Approval workflows
- 20. HP Helion Cloud • Bringing you the benefits of 3 major cloud ecosystems • CF for PaaS • Amazon for cloud burst in Hybrid Cloud • OpenStack for Infrastrcture management
- 21. HP Helion Cloud • OpenStack is the foundation of HP’s Cloud
- 22. HP Helion Cloud • Messaging – Marconi • Database – Trove • IDOL – Integrated Data Operating Layer – Sahara • Docker – OpenStack manage Infrastructure – CF manage Docker applications
- 23. Design Sessions • Nova – Cell enhancement – Scheduler refactor (Mainly for Gantt) • https://etherpad.openstack.org/p/kilo-nova- scheduler-rt – Migrate nova network to neutron – CaaS might be a new incubation project and will not be in nova • Ironic – Create a “standalone” ironic – Simpler to deploy without the rest of OpenStack – Just simpler in general and simpler for the user – Similar with xCat • HEAT – Auto scaling API • https://wiki.openstack.org/wiki/Heat/AutoScaling – HEAT and Container • HEAT is going to integrate with K8S • https://etherpad.openstack.org/p/kilo-heat- containers • https://github.com/rabi/heat-dcoker- docs/blob/master/kilo_design_session_heat_contai ners.pdf – Heat Convergence - resolving any and all questions • Enhance HEAT debug ability • https://wiki.openstack.org/wiki/Heat/ConvergenceD esign • Sahara – How to push it to production – Docker Support – Heat Engine by default – Spark and Hadoop OOB support
- 24. Other Hot Topics • NFV (Network Function Virtualization) • Ceph (Unified Storage Solution) • Just google if you are interested
- 25. OpenStack is calming down • Focusing on usability and debug ability – HEAT – Horizon • Refactor to make architecture more clear – Nova scheduler and Gantt – Horizon • Scalability is on priority – Nova cell – Huawei cascading
- 26. Open Source is eating the software world! Thank You
Editor's Notes
- #4: 峰会的一个概况，大家可以感受下。 5000多人参加 185个session 184个design session 26 个操作相关的session 45个主要的赞助商 43个刚成立的新公司
- #7: Kubernetes是Google开源的容器集群管理系统。它构建于docker技术之上，为容器化的应用提供资源调度、部署运行、服务发现、扩容缩容等整一套功能，本质上可看作是基于容器技术的mini-PaaS平台。
- #8: Atomic是一个用于运行Docker容器的原型系统。是红帽开源的Docker容器管理和部署工具，专门用来托管docker的容器系统。他可以很容易的实现在线升级 类似于coreos
- #11: Openstack的一个项目，主要是 简化应用部署 主要是一个应用控制面板，可以和heat，keystone，nova，trove等交互，帮助用户部署应用。 从右边可以看到，Solum可以支持用户通过拖拽的方式很方便的将一些服务进行排列组合，最终创建一个能够实现某种功能的应用。
- #19: Docker生态系统
- #20: 监控 容量管理和计划 优化资源使用率 Workload的生命周期管理 一些访问权限的控制 混合云的管理 自服务的界面 对用户请求的处理 Quotae的管理 计费 自动化的部署 界面 报表 策略 告警 流程批复
- #21: Helion Cloud可以把openstack，cloudfoundry和amazon连接起来 Os主要是管理infrastructure Cf主要是提供paas服务，帮助用户创建application Amazon提供公有云服务，还可以提供cloud burst功能，本地资源不够的时候，可以到公有云去申请资源
- #22: 私有云 托管云 公有云 企业级云平台 自动化的升级 应用的标准化管理 Cost effective 这个就是说 因为不同公有云的价钱是不一样的 helio cloud能够帮助用户找到最便宜的公有云来部署应用
- #23: Helion cloud上的一些应用