An Intro to Container Orchestrators

Presentation at Shimojo lab seminar

Transcript

1. An Intro to Container Orchestrators
   Shimojo Lab Seminar (Free Talk)
   Nov. 29 2017
   Keichi Takahashi
   

   View Slide

2. What are Container Orchestrators?
   Cluster management software that enables…
   ‣ Launches and manages containers on clusters
   ‣ Scaling and fault recovery of containers
   ‣ Service discovery and load balancing
   Examples include
   ‣ Kubernetes (K8s): Cloud Native Computing Foundation
   ‣ Docker Swarm: Docker
   ‣ Nomad: Hashicorp
   ‣ Marathon: Mesosphere
   2
   

   View Slide

3. Traditional Web Service Architecture
   a.k.a. Three-tier architecture
   3
   Web Server Application Server Database
   Application Server
   Load Balancer Database
   Web Server
   

   View Slide

4. Challenges
   4
   Failure Handling
   Hardware and software
   continuously fail in large-
   scale clusters. Requests
   must be rerouted
   appropriately in case of a
   failure.
   Traffic Surge Deploying Updates
   Incoming traffic often
   fluctuates violently. Server
   resources must be quickly
   scaled out to meet the
   demand.
   Modern web services are
   deployed hundreds or
   thousand times a day.
   Updates must be applied
   without stopping the whole
   service.
   

   View Slide

5. Recent Web Service Architecture
   a.k.a. Microservices architecture
   ‣ Loosely coupled fine-grained services communicating through RPCs
   ‣ Heterogenous tech stack, efficient scaling, high fault tolerance, etc.
   5
   Image
   Conversion
   UI
   Image
   Conversion
   Image
   Conversion
   Image
   Conversion
   Mail
   Image
   Conversion
   Persistency
   Image
   Conversion
   Recommen-
   dation
   Image
   Conversion
   Job Queue
   Image
   Conversion
   Notification
   Image
   Conversion
   Shopping
   Cart
   Image
   Conversion
   Inventory
   Image
   Conversion
   Payment
   

   View Slide

6. Solution: Container Orchestrators
   A software that manages containers distributed across a cluster.
   6
   Container Orchestrator
   Manage
   A A B
   B B C
   B
   B B A A C
   Scale out B
   B
   B B
   A A
   Hardware Crash
   Developer/Operator
   User
   

   View Slide

7. Kubernetes (K8s)
   ‣ The most popular container
   orchestrator today
   ‣ Originally designed by Google,
   donated to the Cloud Native
   Computing Foundation
   ‣ Scales up to 5,000 nodes and
   300,000 containers
   ‣ Used in production at:
   7
   and many more Bloomberg “The Tale of Kubernetes”
   

   View Slide

8. Core Control Loop in Kubernetes
   8
   Diff
   React
   Observe Watch changes in the cluster
   resources
   Compute diff between actual
   cluster state and desired cluster
   state stored in DB
   Reconcile diff by launching
   containers, killing containers,
   etc.
   

   View Slide

9. Architecture of Kubernetes
   9
   kubectl
   kubelet
   kube-proxy
   container
   container
   kube-apiserver
   kube-controller-manager
   kube-scheduler
   kubelet
   kube-proxy
   container
   container
   kubelet
   kube-proxy
   container
   container
   Master
   Node
   Overlay Network
   Overlay Network
   etcd
   

   View Slide

10. Demo
    What I will show here:
    ‣ Start a local Kubernetes cluster
    using minikube
    ‣ Create a replicated deployment
    ‣ Kill a container in the deployment
    ‣ Scale the deployment
    10
    kind: Deployment
    metadata:
    name: nginx
    spec:
    replicas: 3
    template:
    metadata:
    labels:
    app: nginx
    spec:
    containers:
    - name: nginx
    image: nginx
    ports:
    - containerPort: 80
    ---
    kind: Service
    metadata:
    name: nginx-service
    labels:
    name: nginx-service
    spec:
    ports:
    - port: 80
    targetPort: 80
    protocol: TCP
    selector:
    app: nginx
    type: ClusterIP
    

    View Slide

11. The Innovation of Container Orchestrators
    Realized the shift from “how” to “what”
    ‣ From imperative to declarative
    ‣ User requests the desired state, whereas the orchestrator executes
    appropriate procedures to reach the desired state
    Realized the shift from “fortress” to “weeble”
    ‣ Embrace mortality and aim for dynamic equilibrium
    ‣ As long as the cluster can provide services to users as a collective system,
    the life/death of individual nodes or containers doesn’t matter
    11
    

    View Slide

12. The Kubernetes Cluster Whack-a-mole Game
    12
    https://www.youtube.com/watch?v=TyDeZHPJlu8
    

    View Slide

13. Wrap-up
    ‣ A container orchestrator manages containers distributed across a cluster.
    ‣ It enables faster scaling, fault recovery, and deployment.
    ‣ It is changing cluster management from “how” to “what”
    ‣ It is bringing high resiliency to production clusters.
    13
    https://www.cafereo.co.jp/goods/124082
    

    View Slide

14. Further Readings
    The Tale of Kubernetes (Comic!)
    ‣ https://www.bloomberg.com/features/2017-kubernetes/
    Large-scale cluster management at Google with Borg (Paper)
    ‣ https://research.google.com/pubs/pub43438.html
    Scalable Microservices with Kubernetes (MOOC)
    ‣ https://www.udacity.com/course/scalable-microservices-with-
    kubernetes--ud615
    Whack-a-pod: The Kubernetes cluster whack-a-mole game
    ‣ https://medium.com/google-cloud/whack-a-pod-359cbfb61662
    14
    

    View Slide