Virtual Machines versus Containers

[MUSIC] 0:00 In the last stage we took a look at what Docker is and why people use it. 0:04 In this stage we'll look at how Docker can streamline your software development in 0:09 a bid way. 0:13 Long ago, developers had to install all of their apps dependencies on the OS itself 0:15 before they could even begin development work. 0:19 This may work for your purposes, but when it comes time to let a team mate or 0:22 customer use your code, there can be many problems. 0:26 It's not always easy to reproduce what worked on your machine on their machine. 0:29 Then came virtual machines which really save the day. 0:34 VMs essentially emulated entire computers with in a developer's computer, 0:36 VMs were great because they allow developers to install their app's 0:41 dependencies in a virtual machine keeping it isolated from main OS. 0:46 In certain configurations and for 0:51 certain apps, they can even be faster than running on native hardware. 0:52 But this isn't the case most of the time. 0:57 Because a virtual machine has to emulate all the hardware and 0:59 firmware it runs on, it can be slow to start up. 1:02 And VMs are slower to run in most circumstances. 1:06 And because the images contain an entire operating system including packages 1:08 an app may not need, the disk images are big and impractical to distribute. 1:13 Despite these drawbacks, virtual machines made it much easier to build 1:18 reproducible environments for apps to run on, so they exploded in popularity. 1:22 And in 2010, a software project named Vagrant came along, 1:27 that promised to automate the creation of VMs. 1:31 Vagrant allowed developers to build new VMs using a configuration file 1:33 that specified an OS and software packages to install. 1:38 But Vagrant still suffered from the imparent problem of virtual machines, 1:42 distributing applications was still difficult, it wasn't practical to 1:47 pass the VM from developer to developer or developer to customer. 1:51 Anyone who wanted to work on Vagrant VM had 1:55 run the initial bill to create their VM from scratch. 1:58 That was the situation when Docker and the concept of containers arrived in 2013. 2:01 The contents of Docker containers run directly on 2:06 top of the host operating system. 2:09 Unlike a virtual machine where they run on top of a guest operating system. 2:10 Eliminating that guest OS and emulated hardware 2:16 makes Docker containers much more lightweight than virtual machines. 2:19 A Docker still offers many of the same benefits of Virtual Machines, 2:23 because they're isolated in a container, the services running in docker containers 2:27 can't interfere with each other, or with services on the host. 2:31 Because there's no need for a full guest OS, 2:35 docker containers offer many benefits over VMs. 2:37 They start up faster, because you don't have to wait for 2:40 a whole separate OS to load. 2:43 You can re-use systems resources more efficiently. 2:45 A virtual machine, usually has to receive some parts of the host resources and 2:48 can't release them when it's not actively using them. 2:52 Docker containers can free resources they are not using. 2:54 Containers can usually take of the hard role in the host machine and 2:58 ways a VM can. 3:02 For example a VMK usually use the GPU to run codes but a container is up 3:04 can make full use of it, and there's less redundancy with containers. 3:09 Whereas VMs usually need the entire guess OS in order to run successfully, 3:14 containers runs strip down versions of the OS with unnecessary parts removed. 3:18 That means you can run more containers on the same host than you can with VMs. 3:24 There's one limitation you should be aware of. 3:27 Depending on the app you're running a container may not be having faster than 3:31 a virtual machine. 3:34 Containers usually start up faster than VMs though so 3:35 you'll still see a speed benefit from switching. 3:38 Docker's architecture has significant differences from traditional virtual 3:41 machine architecture. 3:45 Docker runs containers on top of light weight virtualized environments without 3:47 fully virtualized in the home machine. 3:51 Because docker doesn't virtualizing entire OS, it can run much quicker and 3:53 be far more affordable than a traditional VM. 3:57 Both virtual machines and containers run on top of a host operating system. 4:00 To run virtual machines, the host OS runs software called a hypervisor. 4:05 A hypervisor manages the life cycle of VMs, starting and stopping them as needed, 4:10 managing their resources and keeping the whole 4:14 setup secure by isolating the VMs from the host OS and from each other. 4:17 An entire guest operating system runs within each virtual machine, and 4:21 your apps dependencies as well as the app itself are run within the guest OS. 4:25 Contrast this with Docker, 4:30 where containers are run by the Docker Daemon software. 4:32 Docker containers omit most of the guest OS except for a few packages. 4:34 Meaning most of the container resources are devoted to your apps dependencies and 4:39 the app itself. 4:43 Docker doesn't have to run the entire guest OS to run an app, 4:44 it simply share the host resources like processors, memory and 4:48 networking devices between each container. 4:51 Even better Docker does an exclusively hold the resources of the host OS. 4:54 When resources are released from a container other Docker containers can use 4:59 those same resources. 5:03 Rather than entire virtual machines, it's simpler to think of containers of 5:05 very heavy processes which the Docker components manage. 5:09 This is what makes Docker a faster and better way to deploy absent services. 5:12 So that's how Docker containers differ from virtual machines. 5:18 Up next we'll look at the details of how Docker works. 5:21