The document discusses the advantages of containerization over traditional hypervisor-based virtualization in cloud datacenters, highlighting the efficiency and resource allocation benefits of containers like Docker. It details how major tech companies have adopted containers, including features of Docker and Docker Swarm for managing container clusters and scheduling. Additionally, it outlines the differences between virtual machines and containers, emphasizing faster boot times and lower resource overhead with container usage.

1. Introduction
Most of cloud datacenters use virtualization which is implemented by
hypervisors such as Xen, KVM or Vsphere.
Container as a good alternative to hypervisor
Most of major IT vendors such as Google, Amazon, and Microsoft and
academic researchers have been interested in containerization of cloud
datacenters to overcome drawbacks of traditional hypervisor based
virtualization[02].
Google: 'EVERYTHING at Google runs in a container‘; Two billion containers a
week
 Docker Container, Linux Container(LxC), Warden Container, Open Vz Container.

2. Container
 “Containerization," an alternative to full machine virtualization in
which an application is encapsulated in a container with its own
operating environment.
 Containerization is an attractive solution that enables developers to
iterate faster.
 It also offers additional benefits that address the overhead associated
with virtual machines, allowing for higher utilization of resources in
the software-defined data center.

3. Virtual Machine Vs Container
Source: https://www.docker.com/
Virtualized Hardware
Operating System
Software
Application
Extra Overhead to Run Machine
Density is Limited (No of VM’s)

4. Virtual Machine Vs Container
Source: https://www.docker.com/
More Density (No of Container’s)
Booting time is fast
Don’t Emulate the whole hardware
Drawback is Same OS

5. Layers of containers
Kernel is same for all
applications
Bootfs Virtual File system to
Docker Images
Layer of Images Readable
Container Writable.
If one VM Image is require
1GB, then 100 VM’s require 100
GB
 If one Container is require
1GB, then 100 Containers
require 1GB(Sharing the Image)
Source: https://www.docker.com/

6. Docker Hub

7. Linux Containers
Linux distribution provide kernel mechanisms such as
oNamespaces: Allows groups of processes to be separated,
preventing them from seeing resources in other groups.
oControl groups: cgroups (abbreviated from control groups) is a
Linux kernel feature that limits, accounts, and isolates the resource
usage (CPU, memory, disk I/O, network, etc.) of a collection of
processes.
Documentation of cgroups-v2 first appeared in Linux kernel 4.5 released on
March 14, 2016

8. Linux Containers

9. 6 Container Technologies
Docker
Docker
Docker Swarm
Linux Container(LxC)
Kubernetes

10. 6.1 Docker
Docker is the world’s leading software container platform.
Docker is an open platform for developers and sysadmins to
build, ship, and run distributed applications, whether on
laptops, data center VMs, or the cloud.

11. 6.1 Docker Vs VM

12. 6.1 Docker Vs VM

13. 7. Docker Swarm
Docker swarm mode is Docker’s solution for managing a cluster of Docker
engines running on multiple machines called SWARM, Where you can
deploy application services which are Docker Container.
Scheduler Swarm comes with a built-in scheduler. It currently provides
basic functionalities, such as scheduling containers based on constraints or
affinity (co-scheduling of containers), persistent storage and multiple
scheduling strategies like binpacking or random.
As per the document of Docker swarm they said, “We plan to add more
features to the built-in scheduler such as rebalancing (in case of host
failure) and global scheduling (schedule containers on every node)”
o Virtual Container ID
o Rebalancing
o Global scheduling
Source: https://github.com/docker/swarm/blob/master/ROADMAP.md Posted May 27,2016

14. 7. Scheduling Using Docker Swarm
Swarm available filters
 Port
 Health
Constraint
Affinity
 Docker Swarm strategies
 Binpack
 Random
Spread

15. 7.1.1 Swarm available Filters
Port
Docker Engine
Docker Engine Docker Engine
Swarm Manager
Nginx: 6379
Nginx: 6379

16. 7.1.2 Swarm available Filters
Health
Docker Engine
Docker Engine Docker Engine
Swarm Manager
Nginx: 6379
Nginx: 6379
Node is Not
Healthy

17. 7.1.3 Swarm available Filters
Constraint
Docker Engine
Docker Engine Docker Engine
Swarm Manager
Nginx: 6379
Nginx: 6379
Constraint==Ubuntu
Fedora
Specific OS
Kernel &
Hardware
Arch Linux Ubuntu

18. 7.1.4 Swarm available Filters
Affinity
Docker Engine
Docker Engine Docker Engine
Swarm Manager
Nginx: 6379
Nginx: 6379
Side By Nginx
Fedora
Attraction
Between
Container
Along With,
Not Along
With
Arch Linux Ubuntu

19. 7.2.1 Docker Swarm strategies
Binpack
Docker Engine
Docker Engine Docker Engine
Swarm Manager
Nginx: 6379
Nginx: 6379
Side By Nginx
Attraction
Between
Container
Along With,
Not Along
With

20. 7.2.2 Docker Swarm strategies
Random
Docker Engine
Docker Engine Docker Engine
Swarm Manager
Nginx: 6379
Nginx: 6379
Side By Nginx
Attraction
Between
Container
Along With,
Not Along
With

21. 7.2.3 Docker Swarm strategies
Spread
Docker Engine
Docker Engine Docker Engine
Swarm Manager
Nginx: 6379
Nginx: 6379
Side By Nginx
Attraction
Between
Container
Along With,
Not Along
With