.NET Developer Days - So many Docker platforms, so little time...
1 like1,062 views
The document discusses various container orchestration platforms, including Docker Swarm, Amazon ECS, Google Container Engine, and Azure Container Service, focusing on features such as scheduling, load balancing, service discovery, and recovery. It addresses frequently asked questions about the need for an orchestration platform and how to choose the right one, along with deployment scenarios and resource constraints. The document also provides insights into using Docker and managing multi-container applications through demos and technical considerations.
.NET Developer Days - So many Docker platforms, so little time...
- 1. So many Docker platforms, so little time… Michele Leroux Bustamante Cofounder / CIO Solliance Microservices / Security Architect @michelebusta
- 3. #1 How do I know if I need an orchestration platform?
- 4. #2 How do I CHOOSE the right approach or platform?
- 5. Considerations Deployment scenarios Single container / instance Classic load balancing Orchestration platforms at scale Features Scheduling and resource constraints Load balancing and networking Service discovery Recovery and self-healing Auto-scale Overall manageability
- 6. In ~ 60 minutes… Docker things Quick review of Docker / container lifecycle Docker-compose, networking, DNS Deploying containers (Azure Container Instance) Platform things Scheduling and constraints (Docker Swarm / Azure Container Service) Orchestration / management UI intro (Docker Enterprise Edition) Load balancing, discovery (DC/OS / Azure Container Service) Recovery, self-healing, auto-scale (Amazon ECS, Kubernetes / Google Container Engine)
- 7. Quick review, Docker things DEMO
- 8. Docker Compose Run multi-container applications Create a service definition file YML (YAML) file “docker-compose.yml” Defines: Container image Ports Networks Volumes Environment variables Configuration
- 9. Image Registry Local Machine Container Registry Push Images Image Repo Latest Version Info Repository tags Promotion tags Tagged Images Images Build Images Code Dockerfile
- 10. Automated Image Builds GitHub / TFS CI Tool Container Registry Build Images Pull Latest Push Images Service Image Repo Latest Version Info Repository tags Promotion tags Dockerfile Checkin Automate or manually execute build Service Repo Tagged Images
- 11. Push images and deploy services DEMO
- 14. What does deployment topology look like?
- 15. Containers and VMs Azure, AWS, Google all support this with IaaS Docker-enabled VMs Load balanced Azure PaaS options App Service with Linux container support Azure Container Instance
- 17. We’ll be ok with a single instance. Containers start fast. NO!
- 18. VM Topologies Single VM Load balanced VMs IaaS Web Apps Load balanced VMs
- 19. VM Topologies Single VM Load balanced VMs IaaS Web Apps Load balanced VMs Azure Container Instance Serverless
- 24. VM Topologies Single VM Load balanced VMs IaaS Web Apps Load balanced VMs Azure Container Instance Serverless Multiple Containers per VM Reverse proxy / router per VM
- 27. Now the fun begins…
- 28. Container Platforms / Considerations Platforms • Amazon ECS • Google Container Engine • Azure Container Service • Mesosphere DC/OS • Kubernetes • Docker Enterprise Edition Infrastructure • Management cluster • Agent node cluster • Proxy / routing/ discovery infrastructure • Docker registry Core Features • Service registration and discovery • Load balancing • Routing • Auto-scaling • Self-healing • Upgrade, rollback, recovery • Versioning
- 29. Platform Master and Agent clusters LEADER Master Nodes / Cluster Agent Nodes
- 30. Scheduling and constraints Docker Swarm
- 31. Docker Swarm
- 32. Scheduling LEADER Master Nodes / Cluster Agent Nodes Schedule services Pull image Run container Service Description
- 33. ### making use of the swarm cluster #Show swarm cluster info $ docker -H tcp://0.0.0.0:2375 info #Show containers $ docker -H tcp://0.0.0.0:2375 ps #Run container on cluster $ docker -H tcp://0.0.0.0:2375 run -d -p 8080:8080 dasblonde/helloworlddocker $ docker -H tcp://0.0.0.0:2375 ps #Run another container, same port (gets different node) $ docker -H tcp://0.0.0.0:2375 run -d -p 8080:8080 dasblonde/helloworlddocker $ docker -H tcp://0.0.0.0:2375 ps
- 34. ###container constraints (port) #Run another container, same port (no more 8080 ports available, error) $ docker -H tcp://0.0.0.0:2375 run -d -p 8080:8080 dasblonde/helloworlddocker $ docker logs --tail 20 containers_swarm_1 #Make request to containers $ curl 10.0.0.4:8080 $ curl 10.0.0.7:8080
- 35. #container constraints (image affinity, memory) #Memory affinity and image affinity $ docker -H tcp://0.0.0.0:2375 run -d -m 3G dasblonde/helloworlddocker $ docker -H tcp://0.0.0.0:2375 info $ docker -H tcp://0.0.0.0:2375 run -d -e affinity:image==dasblonde/helloworlddocker dasblonde/helloworlddocker #Failure due to applied constraints $ docker -H tcp://0.0.0.0:2375 run -d -m 3G -e affinity:image==dasblonde/helloworlddocker dasblonde/helloworlddocker #Node allocation with available memory $ docker -H tcp://0.0.0.0:2375 run -d -m 2G dasblonde/helloworlddocker $ docker -H tcp://0.0.0.0:2375 ps
- 36. Docker Swarm / scheduling and resource constraints DEMO
- 40. Resource Constraints Implicit constraints Physical hardware Technical limitation (port reuse) Explicit constraints CPU, memory, IO Determines node distribution Enables containment of resource usage Multi-tenancy Avoiding noisy neighbor
- 41. Orchestration / Management Native UI Docker Enterprise Edition
- 43. Load balancing, discovery Mesosphere DC/OS / Azure Container Service
- 44. Mesos / Marathon
- 45. DC/OS and Azure Container Service Load balancing / discovery DEMO
- 46. Recovery, self-healing, auto- scale Amazon ECS Kubernetes / Google Container Engine
- 48. Public ALB Route 53 DNS Availability Zone A Availability Zone B Availability Zone C Public Subnet Private Subnet Private Subnet ECS Nodes ECS Nodes Private Subnet ECS NodesEC2 Nodes Public Subnet EC2 Nodes Public Subnet EC2 Nodes Private ALB
- 49. Server Density
- 50. Server Density
- 51. Room for Upgrades, Restarts What about me???
- 52. Adding Nodes What about me???
- 53. Redistribution of Instances What about me???
- 54. Room for New Instances What about me???
- 55. AWS EC2 Container Services (ECS) Self healing, auto scale, health checks DEMO
- 59. Kubernetes / Google Container Engine Job execution DEMO
- 60. Review Docker without an orchestration platform Features of orchestration platforms we care about Load balancing, networking Scheduling, discovery Recovery, self healing Auto scale A look at Docker Swarm / Docker EE Mesosphere DC/OS / Azure Container Service AWS Kubernetes / Google Container Engine
Editor's Notes
- #2: Docker is a key player in the microservices movement and is arguably the leader in containerization technology. That said, there are many ways to “do Docker”. Between the leading cloud providers AWS, Azure, and Google; plus other platform stacks like Docker/Swarm, Apache Mesos – DC/OS, and Kubernetes; it can get confusing. In this session, Michele will bring her customer experiences building solutions across most of these platforms – to provide you with the highlights, the architecture topologies, and some perspective on the way she helps her customers choose the right platform for their cloud, on premise or hybrid solutions.