Allocation of system components over nodes

• Static components (e.g. source code, data)
• Dynamic components (e.g. processes)
• Availability reliability, performance, scalability, security
• System Engineering, System operators

Deployment Models

• Self-managed
  • managed by organization
  • full control
  • high components
  • demand fluctuations
• Cloud
  • third party managed
  • focus on core mission
  • dynamic adaptions
  • lower cost
  • security/privacy
  • vender lock-in

Cloud Computing

Cloud computing is a paradigm for enabling network access to a scalable and elastic pool of shareable physical or virtual resources with self-service provisioning and administration on-demand

• ISO/IEC

A style of computing where scalable and elastic IT-enabled capabilities are provided as a service to multiple customers using Internet technologies

• Gartner

• on-demand access to resource pools
  • applications, computing, storage, networking, development, deployment platforms, business processes
• again third party managed

Characteristics

• On-demand self-service
• rapid elasticity
• broad network access
• measured service
• resource pooling
• multitenancy

Cloud Models

• Public Cloud
  • provider owned
  • subscription based
• Private Cloud
  • company owned
• Community Cloud
  • optimized for sector/group
• hybrid Cloud
  • mixes models

Service Models

• Software as a Service
  • hosted application
• Platform as a Service
  • application development platforms
  • middleware, Server, database
• Infrastructure as a Service
  • hardware Infrastructure

Other Service Models

• Data storage as a service
  • data store capabilities for use by applications
  • API access
• Function as a service
  • cloud based execution of code
  • abstracts processing
  • provider managed
  • scales as needed
• Analytics as a service
  • platform for big data analysis
  • derive actionable insights
• Security as a service
  • applications/services security
    • e.g. authentications, virus detection, antimalware

Virtualization

• enabling technology
• abstracts view on resources
  • application, hardware, storage, network

Hardware Virtualization

• Simulates Hardware expected by Operating System
  • more efficient use of resources
  • sharing configuration/state
• virtual machine-based approach
• container-based approach

Virtual machine

• Hardware Abstraction (Disk, CPU, NIC, ...)
  • Guest Operating System (OS) runs on
• Full Virtualization
  • Guest OS runs unmodified
• Paravirtualization
  • Guest OS modification needed
• Hardware Assisted Virtualization
  • Special VM CPU Calls

Monitor/Hypervisor

• Software layer that runs VMs
  • intercepts instructions

Type 1 Hypervisor

• runs directly on hardware
• E.g. VMWare ESX, Microsoft Hyper-V, KVM
• VMs dedicated Host

Type 2 Hypervisor

• on top of a guest OS
• E.g. VMWare GSX, VirtualBox, QEMU
• Less efficient

Containers

• Lightweight Virtual Machines
  • Linux processes separation
• Operation System Virtualization
  • no guest OS required
  • namespaces
  • control groups (cgroups)
• Example: Docker, rkt

Namespaces

• Sandbox processes: restricts what processes see, use
  • UTS: host/domain
  • Mount: mount points
  • IPC: inter-process communication
  • Cgroup: control group identity
  • PID: set of process ids
  • Network: network subsystems
  • Time: time system

Control Groups (Cgroups)

• Collection of processes with resource usage parameters
  • limit prioritize, meter, control resource usage
  • CPU, memory, network bandwidth

Docker

• container platform
  • portability
• hypervisors for non-linux OS support

Concepts

• Docker Image
  • Packaging of application and environment
  • Layered building
  • Dockerfile
• Docker registry
  • Docker images repository
  • public/Private
  • DockerHub
• Docker Container
  • Container instantiated from an Image

Steps:

1. Developer tells Docker to build and push image
2. Docker builds image
3. Docker pushes image to registry
4. Developer tells Docker on production machine to run image
5. Docker pulls image from registry
6. Docker runs container from image

Container vs Virtual Machines

Virtual Machines:

• run on top of a hypervisor with a full fledged OS
• can run on several applications
• applications from the same VM can share libraries

Containers:

• run on top of the host OS as a process
• runs a single application
• applications from different containers can share libraries

Advantages

• VM
  • Security
    • VM provides complete isolation
    • container can affect kernel
  • Portability
    • VM is portable across any system
    • Container relies on OS
• Container
  • resource efficiency
  • performance
  • agility

Container Orchestration

• automated scheduling, execution, coordination, management of Containers
  • load-balancing
  • discovery
  • failure-handling
  • replication management
• orchestrator: Docker Swarm, Red Hat OpenShift, Kubernetes

Kubernetes

• Open Source Container orchestrator
  • containers deployment, monitoring, management on clusters
  • abstracts distributed infrastructure
  • cloud development platform

Components

• API Server communicates with Control Panel
• Maintains cluster state (etcd)
• Scheduler schedules to workers
• Controller Manager does cluster level functions
• kube-proxy does networking
• Kubelet interacts with API Server
• Controller Runtime runs containers

Kubernetes Objects

• entities provided by Kubernetes
  • Pods
  • Replicasets
  • Deployments
  • Services
  • Volumes
• YAML/JSON definition

Pods

• Basic building block
• Groups co-located Containers
• Managed as unit

Clusters: