Shulou(Shulou.com)05/31 Report--

This article introduces the relevant knowledge of "what is the concept of Docker container". In the operation of practical cases, many people will encounter such a dilemma, so let the editor lead you to learn how to deal with these situations. I hope you can read it carefully and be able to achieve something!

If you don't play virtual games, of course you'll be happier.

Docker vs. virtual machine (vm):

The virtual machine runs on the virtual hardware and the application runs on the virtual machine kernel. Docker daemon is a process on the host, and the application is just a child process of docker daemon. In other words, the application runs directly on the host kernel.

Virtual machines need special hardware virtualization technology support, so they can only run on physical machines. Docker does not have hardware virtualization, so it can run in physical machines, virtual machines, or even docker containers (nested).

Because there is no hardware virtualization and the overhead of running an extra Linux kernel, applications run lighter and faster on docker than on virtual machines.

So what's the difference between using docker and running a program directly on the host? The answer is that the docker container uses cgroup namespaces to isolate resources such as CPU, memory, network, and file systems. Students who know chroot can think of docker as a more elegant chroot.

Speaking of isolation, where is the file system for the docker container? The answer is mirroring.

Docker Mirror

The image describes the initial file system on which the docker container runs and contains all the dependencies needed to run the application. It can be a complete operating system, or it can contain only the smallest collection of bin/lib files required by the application.

One question: assuming that the file system size of one docker container is 10GiB, how much disk space does it take to create 10 containers?

100 GiB? No, it's only 10 GiB! Because docker images and containers have a hierarchical file system structure, each container contains a thin layer of writable layer, and the read-only part is shared.

Docker image storage engine has many implementations such as aufs, devicemapper, overlay and so on.

The image is read-only, and the container is created only by creating a writable layer on top of the image, without copying the entire file system, so millisecond creation can be achieved.

The mirror is stored in the mirror hub. Each mirror has a tag, such as registry.hub.docker.com/library/ubuntu:14.04. The full tag of the image not only contains the name of the image, but also indicates where the image comes from and where it is going, just like a URL. Yes, registry.hub.docker.com is the address of the official docker image hub. When using an official library image, the prefix can be omitted. For example, the above image tag can be abbreviated to ubuntu:14.04.

Docker is created to run applications

With an image, how to create a container? Docker does not load kernel (unlike vm) or execute init (unlike vm and lxc). "what's the point of running a kernel or init if you don't execute the application?" Docker said. The docker container is created to run the application. To create a docker container, you must specify the image tag and the command line to start the application (or the default command line is set for the image).

So, the command line to create a docker container is as follows:

Sudo docker create-ti-- name test ubuntu:14.04 bash

This is just a container that has not been run yet. To run the container, execute:

Sudo docker start test

Usually we merge these two commands into one, namely docker run:

Sudo docker run-dti-- name test ubuntu:14.04 bash

On the host, take a look at the process tree of the bash command we just ran:

$pstree-sa 21811systemd-- switched-root-- system-- deserialize 21 └─ docker daemon... └─ bash

Yes, it is just a child process of docker daemon.

Application is the main process of docker

The application of attach to the container. In this case, the bash command line:

Sudo docker attach test

Ps-ef take a look:

$sudo docker attach testroot@47c90c3dcbfa:/# ps-efUID PID PPID C STIME TTY TIME CMDroot 1 00 17:05? 00:00:00 bashroot 18 1 0 17:11? 00:00:00 ps-ef

With ps itself, there are only two processes in the container. The PID of bash in the container is 1, which means the main process of the container. After the main process exits, the container automatically exits. Again, the container is created to run the application, and the application has exited, so what's the use of keeping the container?

Stop the container manually:

Sudo docker stop test

Delete the container:

Sudo docker rm test

Note: after deleting the container, the writable layer file system created on the image will also be deleted, so logs, configurations, data and other files that need to be persisted need to be attached to external storage.

How light is docker?

As the saying goes, when predecessors plant trees and future generations enjoy the cool, we usually create an application image based on an existing image. To demonstrate how light the image is, we can create a mirror starting at 0.

Create a folder hello-lark, enter the folder, and add a Dockerfile content as follows:

FROM scratchCOPY Dockerfile hello.c hello /

Dockerfile describes how to create an image:

FROM scratch means starting at 0, that is, creating a mirror from a blank file system.

COPY Dockerfile hello.c hello / means to copy the local Dockerfile, hello.c and hello files to the root directory in the image.

Hello.c and hello are our application source code and compiled binaries.

The contents of the hello.c file are as follows:

# include int main (int argc, char* argv []) {printf ("Hello, Lark!\ n"); return 0;}

Static compilation:

Gcc-static hello.c-o hello

Execute docker build to create an image:

$sudo docker build-t hello-lark:1.0 .Sending build context to Docker daemon 881.2 kBStep 1: FROM scratch-> Step 2: COPY Dockerfile hello.c hello /-> b52b3bd2799cRemoving intermediate container bc5dcd332026Successfully built b52b3bd2799c

The images are packaged and stored in the local warehouse, and the physical files cannot be seen directly. However, it can be exported as a tar package:

Sudo docker save-o.. / hello-lark.tar hello-lark:1.0

See how big it is:

$ll-h.. / hello-lark.tar-rw-r--r-- 1 root root 870K Apr 12 13:41.. / hello-lark.tar

It is only 870K, which is about the size of the binary file hello.

How fast is docker?

Can this mirror run? Let's run it right away and take a look:

$time sudo docker run-- name hello--rm hello-lark:1.0 / helloHello, Larkhorse real 0m0.486suser 0m0.078ssys 0m0.016s

As you can see, this command goes through four processes: create, run, stop, and destroy the container, and it takes less than 0.5 seconds. -- rm means that the container is automatically deleted after the container ends.

If you are just creating:

$time sudo docker create-- name hello hello-lark:1.0 / hello16449a07a0f1377336a879f8136ec7369e07520da1f4a7516128b8d3ba314008real 0m0.075suser 0m0.064ssys 0m0.017s

Less than 0.1 second. Millisecond creation is by no means a lie.

Of course, just hello world is useless, we hit a busybox, the image size is only 2MB, but we already have a basic linux shell environment.

Sudo docker run-ti-the benefits of rm lark-box:1.0 shdocker

To put it simply: standardized delivery, micro-service orchestration, improve resource utilization.

Calibrated delivery

Docker packages the application and all its dependencies into an image, including binaries (including underlying libraries), static configuration files, environment variables, and so on. Stripping off the dependence of the application on the operating system and environment, loosely coupled. You only need to pull the image and start the container to complete the application deployment. Create and destroy containers in milliseconds, so that you can achieve rapid deployment, rapid migration, rapid capacity expansion and reduction, and quick rollback with one click (only depends on the startup time of the application). Docker image establishes application delivery standards, developers have complete control over the application and its running environment, and effectively avoid a variety of environmental problems.

Micro service choreography

When deploying multiple applications on a single machine, the applications are completely decoupled, and the orchestration can be deployed arbitrarily, which perfectly supports the needs of micro-service orchestration. When multiple C applications are mixed, docker realizes C dependency isolation and avoids dependency conflicts.

Improve the utilization rate of resources

Docker is a lightweight solution that does not do virtualization and does not run redundant kernel and init processes, which can effectively improve resource utilization.

This is the end of "what is the concept of Docker Container". Thank you for reading. If you want to know more about the industry, you can follow the website, the editor will output more high-quality practical articles for you!

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.