Container Landscape in 2019
- 1. Anusha Ragunathan Senior Software Engineer, Docker Women Of Silicon Valley, May 3, 2019 Container Landscape in 2019
- 2. ● Sr Software Engineer at Docker ● open source maintainer of docker/moby ● Mom of 2 girls >whoami
- 3. ● Containers: Today and Beyond ● Container Day 2 operations ● Importance of open APIs Agenda
- 5. ● Mechanism to run software. ● Provides repeatability ● Provides isolation ● Distributed using a registry What are containers?
- 7. ● Traditional Use cases: ○ Consolidation ○ Cloud Native Applications ● Emerging Use cases: ○ Edge ○ Serverless Motivation: Why move to containers
- 8. ● Containers reduce your infra cost ○ Under utilization of VM instances. You get higher cloud bills, but paying for infrastructure costs that are not being used. ● When you move to containers, this is a common denominator across cloud providers. Traditional use-case: Consolidation
- 9. Efficiency. 5X over VMs 8 VMs 40 containers
- 10. ● New wave of applications that are rooted in containerization and are built as microservices. For these, containers are a given ● Cloud Native Computing Foundation (CNCF) Landscape is rich ● Ecosystem has everything needed to have a working production grade solution for your problems ● Microservices offer rapid deployment and scalability Traditional use-case: Cloud Native
- 12. ● Categories that are worth pointing out: ➢ App definition and development - Developer ➢ Container Orchestration and Management - Enterprise ➢ Container Runtime: Runtimes, Storage and Network - Enterprise ➢ Container Provisioning - Enterprise ➢ Monitoring, logging and tracing - Developer + Enterprise Cloud Native Landscape
- 13. ● Edge devices and IoT world are adopting containers and COs ● Containers are well suited for the small-footprint, low-latency environments ● Example: ○ Docker can build, ship and run on arm64 ○ AWS EC2 A1 instances on the cloud ○ Lightweight Kubernetes distribution such as Rancher’s k3s for the edge Emerging use-case : Edge
- 14. ● Serverless technologies being built on containers. ● Easier to deploy and no vendor lock-in. ● Example: ○ OpenFaas ○ Apache OpenWhisk Emerging use-case : Serverless
- 16. Sock shop setup ○ Sockshop is a demo e-commerce application designed as microservices. ○ Written using ■ Spring Boot ■ Go Kit ■ Node.js ○ Packaged in Docker containers ○ Deployed on Kubernetes
- 18. Container - Day 2 ops Orchestration, Storage and Observability
- 19. Container Orchestration ➢ Container lifecycle management ➢ Placement across cluster ➢ Elasticity and scale ➢ Service discovery, load balancing and HA proxy ➢ Standardized on Kubernetes
- 20. Container Orchestration ➢ Dev and test ○ Docker Desktop with k8s integration ○ Ubuntu microk8s ➢ Managed Kubernetes in the cloud ○ AKS, EKS, GKE ➢ Kubernetes on Edge ○ Docker Edge, k3s ➢ Several K8s distros
- 21. Container Storage Containers started out to be stateless. Microservices: ➢ Are short-lived on a host and orchestrator tends to move apps across the cluster. ➢ Have containers that require ○ fast-provisioning of storage volumes ○ repeat attach detach of provisioning volumes to cluster nodes ○ mounting these volumes into the microservice. ➢ Hyper scale and different on-premises and cloud environments that containers can live in make this even harder
- 22. Container Storage Solutions: ➢ Open standard spec in Container Orchestrator for Storage ➢ Container Storage Interface (CSI) is a standard to provide a pluggable framework for storage providers. This helps with seamless and consistent deployment of storage solutions.
- 23. Challenges: ➢ Container-based microservices create flexible environments to develop, distribute and deploy ➢ However, once the services are deployed, visibility of interactions between services, as well as insights into cluster infrastructure, becomes increasingly complex. Container Observability
- 24. Solutions: ➢ DevOps teams use Application Performance Management (APM) tools to view dashboards to debug complex microservice-based interactions ➢ Open specification around distributed tracing called OpenTracing. Here applications instrument their code to spew out traces while running. These traces can be collected by analytics software, which helps derive inferences from the collected traces ➢ APM tools are becoming quintessential to DevOps Container Observability
- 25. Why open APIs?
- 26. Why open APIs? ● Build-Ship-Run lifecycle of containers is OPEN ○ Build : App building specifications are open ○ Ship: Container distribution specifications are open ○ Run: Container Runtimes and specifications are open ○ Orchestrators: Cloud providers, Kubernetes, Mesos, etc ○ Avoid vendor lock-ins. Can move from one provider to another easily.
- 27. Takeaways ➢ Containers are here to stay ➢ CNCF Landscape is huge and expanding ➢ Open source and open APIs for the win!
- 28. Thank you!