Skip to content

Latest commit

 

History

History
864 lines (654 loc) · 35.9 KB

architecture.md

File metadata and controls

864 lines (654 loc) · 35.9 KB
Raw

Kata Containers Architecture

Overview

Kata Containers is an open source community working to build a secure container runtime with lightweight virtual machines (VM's) that feel and perform like standard Linux containers, but provide stronger workload isolation using hardware virtualization technology as a second layer of defence.

Kata Containers runs on multiple architectures and supports multiple hypervisors.

This document is a summary of the Kata Containers architecture.

Virtualization

For details on how Kata Containers maps container concepts to VM technologies, and how this is realized in the multiple hypervisors and VMMs that Kata supports see the virtualization documentation.

Compatibility

The Kata Containers runtime is compatible with the OCI runtime specification and therefore works seamlessly with the Kubernetes Container Runtime Interface (CRI) through the CRI-O and containerd implementations.

Kata Containers provides a "shimv2" compatible runtime.

Shim v2 architecture

The Kata Containers runtime is shim v2 ("shimv2") compatible. This section explains what this means.

History

In the old Kata 1.x architecture, the Kata runtime was an executable called kata-runtime. The container manager called this executable multiple times when creating each container. Each time the runtime was called a different OCI command-line verb was provided. This architecture was simple, but not well suited to creating VM based containers due to the issue of handling state between calls. Additionally, the architecture suffered from performance issues related to continually having to spawn new instances of the runtime binary, and Kata shim and Kata proxy processes for systems that did not provide VSOCK.

An improved architecture

The containerd runtime shimv2 architecture or shim API architecture resolves the issues with the old architecture by defining a set of shimv2 APIs that a compatible runtime implementation must supply. Rather than calling the runtime binary multiple times for each new container, the shimv2 architecture runs a single instance of the runtime binary (for any number of containers). This improves performance and resolves the state handling issue.

The shimv2 API is similar to the OCI runtime API in terms of the way the container lifecycle is split into different verbs. Rather than calling the runtime multiple times, the container manager creates a socket and passes it to the shimv2 runtime. The socket is a bi-directional communication channel that uses a gRPC based protocol to allow the container manager to send API calls to the runtime, which returns the result to the container manager using the same channel.

The shimv2 architecture allows running several containers per VM to support container engines that require multiple containers running inside a pod.

With the new architecture Kubernetes can launch both Pod and OCI compatible containers with a single runtime shim per Pod, rather than 2N+1 shims. No stand alone kata-proxy process is required, even if VSOCK is not available.

Architectural comparison

Kata version Kata Runtime process calls Kata shim processes Kata proxy processes (if no VSOCK)
1.x multiple per container 1 per container connection 1
2.x 1 per VM (hosting any number of containers) 0 0

Notes:

  • A single VM can host one or more containers.

  • The "Kata shim processes" column refers to the old Kata shim (kata-shim binary), not the new shimv2 runtime instance (containerd-shim-kata-v2 binary).

The diagram below shows how the original architecture was simplified with the advent of shimv2.

Kubernetes integration with shimv2

Root filesystem

This document uses the term rootfs to refer to a root filesystem which is mounted as the top-level directory ("/") and often referred to as slash.

It is important to understand this term since the overall system uses multiple different rootfs's (as explained in the Environments section.

Example command

The following containerd command creates a container. It is referred to throughout this document to help explain various points:

$ sudo ctr run --runtime "io.containerd.kata.v2" --rm -t "quay.io/libpod/ubuntu:latest" foo sh

This command requests that containerd:

  • Create a container (ctr run).

  • Use the Kata shimv2 runtime (--runtime "io.containerd.kata.v2").

  • Delete the container when it exits (--rm).

  • Attach the container to the user's terminal (-t).

  • Use the Ubuntu Linux container image to create the container rootfs that will become the container environment (quay.io/libpod/ubuntu:latest).

  • Create the container with the name "foo".

  • Run the sh(1) command in the Ubuntu rootfs based container environment.

    The command specified here is referred to as the workload.

Note:

For the purposes of this document and to keep explanations simpler, we assume the user is running this command in the host environment.

Container image

In the example command the user has specified the type of container they wish to run via the container image name: ubuntu. This image name corresponds to a container image that can be used to create a container with an Ubuntu Linux environment. Hence, in our example, the sh(1) command will be run inside a container which has an Ubuntu rootfs.

Note:

The term container image is confusing since the image in question is not a container: it is simply a set of files (an image) that can be used to create a container. The term container template would be more accurate but the term container image is commonly used so this document uses the standard term.

For the purposes of this document, the most important part of the example command line is the container image the user has requested. Normally, the container manager will pull (download) a container image from a remote site and store a copy locally. This local container image is used by the container manager to create an OCI bundle which will form the environment the container will run in. After creating the OCI bundle, the container manager launches a runtime which will create the container using the provided OCI bundle.

OCI bundle

To understand what follows, it is important to know at a high level how an OCI (Open Containers Initiative) compatible container is created.

An OCI compatible container is created by taking a container image and converting the embedded rootfs into an OCI rootfs bundle, or more simply, an OCI bundle.

An OCI bundle is a tar(1) archive normally created by a container manager which is passed to an OCI runtime which converts it into a full container rootfs. The bundle contains two assets:

  • A container image rootfs

    This is simply a directory of files that will be used to represent the rootfs for the container.

    For the example command, the directory will contain the files necessary to create a minimal Ubuntu root filesystem.

  • An OCI configuration file

    This is a JSON file called config.json.

    The container manager will create this file so that:

    • The root.path value is set to the full path of the specified container rootfs.

      In the example this value will be ubuntu.

    • The process.args array specifies the list of commands the user wishes to run. This is known as the workload.

      In the example the workload is sh(1).

Workload

The workload is the command the user requested to run in the container and is specified in the OCI bundle's configuration file.

In our example, the workload is the sh(1) command.

Workload root filesystem

For details of how the runtime makes the container image chosen by the user available to the workload process, see the Container creation and storage sections.

Note that the workload is isolated from the guest VM environment by its surrounding container environment. The guest VM environment where the container runs in is also isolated from the outer host environment where the container manager runs.

System overview

Environments

The following terminology is used to describe the different or environments (or contexts) various processes run in. It is necessary to study this table closely to make sense of what follows:

Type Name Virtualized Containerized rootfs Rootfs device type Mount type Description
Host Host no [1] no Host specific Host specific Host specific The environment provided by a standard, physical non virtualized system.
VM root Guest VM yes no rootfs inside the guest image Hypervisor specific [2] ext4 The first (or top) level VM environment created on a host system.
VM container root Container yes yes rootfs type requested by user (ubuntu in the example) kataShared virtio FS The first (or top) level container environment created inside the VM. Based on the OCI bundle.

Key:

  • [1]: For simplicity, this document assumes the host environment runs on physical hardware.

  • [2]: See the DAX section.

Notes:

  • The word "root" is used to mean top level here in a similar manner to the term rootfs.

  • The term "first level" prefix used above is important since it implies that it is possible to create multi level systems. However, they do not form part of a standard Kata Containers environment so will not be considered in this document.

The reasons for containerizing the workload inside the VM are:

  • Isolates the workload entirely from the VM environment.
  • Provides better isolation between containers in a pod.
  • Allows the workload to be managed and monitored through its cgroup confinement.

Container creation

The steps below show at a high level how a Kata Containers container is created using the containerd container manager:

  1. The user requests the creation of a container by running a command like the example command.

  2. The container manager daemon runs a single instance of the Kata runtime.

  3. The Kata runtime loads its configuration file.

  4. The container manager calls a set of shimv2 API functions on the runtime.

  5. The Kata runtime launches the configured hypervisor.

  6. The hypervisor creates and starts (boots) a VM using the guest assets:

  7. The agent is started as part of the VM boot.

  8. The runtime calls the agent's CreateSandbox API to request the agent create a container:

    1. The agent creates a container environment in the container specific directory that contains the container rootfs.

      The container environment hosts the workload in the container rootfs directory.

    2. The agent spawns the workload inside the container environment.

    Notes:

    • The container environment created by the agent is equivalent to a container environment created by the runc OCI runtime; Linux cgroups and namespaces are created inside the VM by the guest kernel to isolate the workload from the VM environment the container is created in. See the Environments section for an explanation of why this is done.

    • See the guest image section for details of exactly how the agent is started.

  9. The container manager returns control of the container to the user running the ctr command.

Note:

At this point, the container is running and:

Further details of these steps are provided in the sections below.

Container shutdown

There are two possible ways for the container environment to be terminated:

  • When the workload exits.

    This is the standard, or graceful shutdown method.

  • When the container manager forces the container to be deleted.

Workload exit

The agent will detect when the workload process exits, capture its exit status (see wait(2)) and return that value to the runtime by specifying it as the response to the WaitProcess agent API call made by the runtime.

The runtime then passes the value back to the container manager by the Wait shimv2 API call.

Once the workload has fully exited, the VM is no longer needed and the runtime cleans up the environment (which includes terminating the hypervisor process).

Note:

When agent tracing is enabled, the shutdown behaviour is different.

Container manager requested shutdown

If the container manager requests the container be deleted, the runtime will signal the agent by sending it a DestroySandbox ttRPC API request.

Guest assets

Kata Containers creates a VM in which to run one or more containers. It does this by launching a hypervisor to create the VM. The hypervisor needs two assets for this task: a Linux kernel and a small root filesystem image to boot the VM.

Guest kernel

The guest kernel is passed to the hypervisor and used to boot the VM. The default kernel provided in Kata Containers is highly optimized for kernel boot time and minimal memory footprint, providing only those services required by a container workload. It is based on the latest Linux LTS (Long Term Support) kernel.

Guest image

The hypervisor uses an image file which provides a minimal root filesystem used by the guest kernel to boot the VM and host the Kata Container. Kata Containers supports both initrd and rootfs based minimal guest images. The default packages provide both an image and an initrd, both of which are created using the osbuilder tool.

Notes:

  • Although initrd and rootfs based images are supported, not all hypervisors support both types of image.

  • The guest image is unrelated to the image used in a container workload.

    For example, if a user creates a container that runs a shell in a BusyBox image, they will run that shell in a BusyBox environment. However, the guest image running inside the VM that is used to host that BusyBox image could be running Clear Linux, Ubuntu, Fedora or any other distribution potentially.

    The osbuilder tool provides configurations for various common Linux distributions which can be built into either initrd or rootfs guest images.

  • If you are using a packaged version of Kata Containers, you can see image details by running the kata-collect-data.sh script as root and looking at the "Image details" section of the output.

Root filesystem image

The default packaged rootfs image, sometimes referred to as the mini O/S, is a highly optimized container bootstrap system.

If this image type is configured, when the user runs the example command:

  • The runtime will launch the configured hypervisor.
  • The hypervisor will boot the mini-OS image using the guest kernel.
  • The kernel will start the init daemon as PID 1 (systemd) inside the VM root environment.
  • systemd, running inside the mini-OS context, will launch the agent in the root context of the VM.
  • The agent will create a new container environment, setting its root filesystem to that requested by the user (Ubuntu in the example).
  • The agent will then execute the command (sh(1) in the example) inside the new container.

The table below summarises the default mini O/S showing the environments that are created, the services running in those environments (for all platforms) and the root filesystem used by each service:

Process Environment systemd service? rootfs User accessible Notes
systemd VM root n/a VM guest image debug console The init daemon, running as PID 1
Agent VM root yes VM guest image debug console Runs as a systemd service
chronyd VM root yes VM guest image debug console Used to synchronise the time with the host
container workload (sh(1) in the example) VM container no User specified (Ubuntu in the example) exec command Managed by the agent

See also the process overview.

Notes:

  • The "User accessible" column shows how an administrator can access the environment.

  • The container workload is running inside a full container environment which itself is running within a VM environment.

  • See the configuration files for the osbuilder tool for details of the default distribution for platforms other than Intel x86_64.

Initrd image

The initrd image is a compressed cpio(1) archive, created from a rootfs which is loaded into memory and used as part of the Linux startup process. During startup, the kernel unpacks it into a special instance of a tmpfs mount that becomes the initial root filesystem.

If this image type is configured, when the user runs the example command:

  • The runtime will launch the configured hypervisor.
  • The hypervisor will boot the mini-OS image using the guest kernel.
  • The kernel will start the init daemon as PID 1 (the agent) inside the VM root environment.
  • The agent will create a new container environment, setting its root filesystem to that requested by the user (ubuntu in the example).
  • The agent will then execute the command (sh(1) in the example) inside the new container.

The table below summarises the default mini O/S showing the environments that are created, the processes running in those environments (for all platforms) and the root filesystem used by each service:

Process Environment rootfs User accessible Notes
Agent VM root VM guest image debug console Runs as the init daemon (PID 1)
container workload VM container User specified (Ubuntu in this example) exec command Managed by the agent

Notes:

  • The "User accessible" column shows how an administrator can access the environment.

  • It is possible to use a standard init daemon such as systemd with an initrd image if this is desirable.

See also the process overview.

Image summary

Image type Default distro Init daemon Reason Notes
image Clear Linux (for x86_64 systems) systemd Minimal and highly optimized systemd offers flexibility
initrd Alpine Linux Kata agent (as no systemd support) Security hardened and tiny C library

See also:

  • The osbuilder tool

    This is used to build all default image types.

  • The versions database

    The default-image-name and default-initrd-name options specify the default distributions for each image type.

Hypervisor

The hypervisor specified in the configuration file creates a VM to host the agent and the workload inside the container environment.

Note:

The hypervisor process runs inside an environment slightly different to the host environment:

  • It is run in a different cgroup environment to the host.
  • It is given a separate network namespace from the host.
  • If the OCI configuration specifies a SELinux label, the hypervisor process will run with that label (not the workload running inside the hypervisor's VM).

Agent

The Kata Containers agent (kata-agent), written in the Rust programming language, is a long running process that runs inside the VM. It acts as the supervisor for managing the containers and the workload running within those containers. Only a single agent process is run for each VM created.

Agent communications protocol

The agent communicates with the other Kata components (primarily the runtime) using a ttRPC based protocol.

Note:

If you wish to learn more about this protocol, a practical way to do so is to experiment with the agent control tool on a test system. This tool is for test and development purposes only and can send arbitrary ttRPC agent API commands to the agent.

Runtime

The Kata Containers runtime (the containerd-shim-kata-v2 binary) is a shimv2 compatible runtime.

Note:

The Kata Containers runtime is sometimes referred to as the Kata shim. Both terms are correct since the containerd-shim-kata-v2 is a container runtime, and that runtime implements the containerd shim v2 API.

The runtime makes heavy use of the virtcontainers package, which provides a generic, runtime-specification agnostic, hardware-virtualized containers library.

The runtime is responsible for starting the hypervisor and it's VM, and communicating with the agent using a ttRPC based protocol over a VSOCK socket that provides a communications link between the VM and the host.

This protocol allows the runtime to send container management commands to the agent. The protocol is also used to carry the standard I/O streams (stdout, stderr, stdin) between the containers and container managers (such as CRI-O or containerd).

Utility program

The kata-runtime binary is a utility program that provides administrative commands to manipulate and query a Kata Containers installation.

Note:

In Kata 1.x, this program also acted as the main runtime, but this is no longer required due to the improved shimv2 architecture.

exec command

The exec command allows an administrator or developer to enter the VM root environment which is not accessible by the container workload.

See the developer guide for further details.

Configuration

See the configuration file details.

The configuration file is also used to enable runtime debug output.

Process overview

The table below shows an example of the main processes running in the different environments when a Kata Container is created with containerd using our example command:

Description Host VM root environment VM container environment
Container manager containerd
Kata Containers runtime, virtiofsd, hypervisor agent
User workload ubuntu sh

Networking

Containers will typically live in their own, possibly shared, networking namespace. At some point in a container lifecycle, container engines will set up that namespace to add the container to a network which is isolated from the host network, but which is shared between containers

In order to do so, container engines will usually add one end of a virtual ethernet (veth) pair into the container networking namespace. The other end of the veth pair is added to the host networking namespace.

This is a very namespace-centric approach as many hypervisors or VM Managers (VMMs) such as virt-manager cannot handle veth interfaces. Typically, TAP interfaces are created for VM connectivity.

To overcome incompatibility between typical container engines expectations and virtual machines, Kata Containers networking transparently connects veth interfaces with TAP ones using Traffic Control:

Kata Containers networking

With a TC filter in place, a redirection is created between the container network and the virtual machine. As an example, the CNI may create a device, eth0, in the container's network namespace, which is a VETH device. Kata Containers will create a tap device for the VM, tap0_kata, and setup a TC redirection filter to mirror traffic from eth0's ingress to tap0_kata's egress, and a second to mirror traffic from tap0_kata's ingress to eth0's egress.

Kata Containers maintains support for MACVTAP, which was an earlier implementation used in Kata. TC-filter is the default because it allows for simpler configuration, better CNI plugin compatibility, and performance on par with MACVTAP.

Kata Containers has deprecated support for bridge due to lacking performance relative to TC-filter and MACVTAP.

Kata Containers supports both CNM and CNI for networking management.

Network Hotplug

Kata Containers has developed a set of network sub-commands and APIs to add, list and remove a guest network endpoint and to manipulate the guest route table.

The following diagram illustrates the Kata Containers network hotplug workflow.

Network Hotplug

Storage

virtio SCSI

If a block-based graph driver is configured, virtio-scsi is used to share the workload image (such as busybox:latest) into the container's environment inside the VM.

virtio FS

If a block-based graph driver is not configured, a virtio-fs (VIRTIO) overlay filesystem mount point is used to share the workload image instead. The agent uses this mount point as the root filesystem for the container processes.

For virtio-fs, the runtime starts one virtiofsd daemon (that runs in the host context) for each VM created.

Devicemapper

The devicemapper snapshotter is a special case. The snapshotter uses dedicated block devices rather than formatted filesystems, and operates at the block level rather than the file level. This knowledge is used to directly use the underlying block device instead of the overlay file system for the container root file system. The block device maps to the top read-write layer for the overlay. This approach gives much better I/O performance compared to using virtio-fs to share the container file system.

Hot plug and unplug

Kata Containers has the ability to hot plug add and hot plug remove block devices. This makes it possible to use block devices for containers started after the VM has been launched.

Users can check to see if the container uses the devicemapper block device as its rootfs by calling mount(8) within the container. If the devicemapper block device is used, the root filesystem (/) will be mounted from /dev/vda. Users can disable direct mounting of the underlying block device through the runtime configuration.

Kubernetes support

Kubernetes, or K8s, is a popular open source container orchestration engine. In Kubernetes, a set of containers sharing resources such as networking, storage, mount, PID, etc. is called a pod.

A node can have multiple pods, but at a minimum, a node within a Kubernetes cluster only needs to run a container runtime and a container agent (called a Kubelet).

Kata Containers represents a Kubelet pod as a VM.

A Kubernetes cluster runs a control plane where a scheduler (typically running on a dedicated master node) calls into a compute Kubelet. This Kubelet instance is responsible for managing the lifecycle of pods within the nodes and eventually relies on a container runtime to handle execution. The Kubelet architecture decouples lifecycle management from container execution through a dedicated gRPC based Container Runtime Interface (CRI).

In other words, a Kubelet is a CRI client and expects a CRI implementation to handle the server side of the interface. CRI-O and containerd are CRI implementations that rely on OCI compatible runtimes for managing container instances.

Kata Containers is an officially supported CRI-O and containerd runtime. Refer to the following guides on how to set up Kata Containers with Kubernetes:

OCI annotations

In order for the Kata Containers runtime (or any VM based OCI compatible runtime) to be able to understand if it needs to create a full VM or if it has to create a new container inside an existing pod's VM, CRI-O adds specific annotations to the OCI configuration file (config.json) which is passed to the OCI compatible runtime.

Before calling its runtime, CRI-O will always add a io.kubernetes.cri-o.ContainerType annotation to the config.json configuration file it produces from the Kubelet CRI request. The io.kubernetes.cri-o.ContainerType annotation can either be set to sandbox or container. Kata Containers will then use this annotation to decide if it needs to respectively create a virtual machine or a container inside a virtual machine associated with a Kubernetes pod:

Annotation value Kata VM created? Kata container created?
sandbox yes yes (inside new VM)
container no yes (in existing VM)

Mixing VM based and namespace based runtimes

Note: Since Kubernetes 1.12, the Kubernetes RuntimeClass has been supported and the user can specify runtime without the non-standardized annotations.

With RuntimeClass, users can define Kata Containers as a RuntimeClass and then explicitly specify that a pod must be created as a Kata Containers pod. For details, please refer to How to use Kata Containers and containerd.

Tracing

The tracing document provides details on the tracing architecture.

Appendices

DAX

Kata Containers utilizes the Linux kernel DAX (Direct Access filesystem) feature to efficiently map the guest image in the host environment into the guest VM environment to become the VM's rootfs.

If the configured hypervisor is set to either QEMU or Cloud Hypervisor, DAX is used with the feature shown in the table below:

Hypervisor Feature used rootfs device type
Cloud Hypervisor (CH) dax FsConfig configuration option PMEM (emulated Persistent Memory device)
QEMU NVDIMM memory device with a memory file backend NVDIMM (emulated Non-Volatile Dual In-line Memory Module device)

The features in the table above are equivalent in that they provide a memory-mapped virtual device which is used to DAX map the VM's rootfs into the VM guest memory address space.

The VM is then booted, specifying the root= kernel parameter to make the guest kernel use the appropriate emulated device as its rootfs.

DAX advantages

Mapping files using DAX provides a number of benefits over more traditional VM file and device mapping mechanisms:

  • Mapping as a direct access device allows the guest to directly access the host memory pages (such as via Execute In Place (XIP)), bypassing the guest kernel's page cache. This zero copy provides both time and space optimizations.

  • Mapping as a direct access device inside the VM allows pages from the host to be demand loaded using page faults, rather than having to make requests via a virtualized device (causing expensive VM exits/hypercalls), thus providing a speed optimization.

  • Utilizing mmap(2)'s MAP_SHARED shared memory option on the host allows the host to efficiently share pages.

DAX

For further details of the use of NVDIMM with QEMU, see the QEMU project documentation.

Agent control tool

The agent control tool is a test and development tool that can be used to learn more about a Kata Containers system.

Terminology

See the project glossary.