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

This article mainly introduces "what is the real-time migration of docker containers". In the daily operation, I believe many people have doubts about what the real-time migration of docker containers is. The editor consulted all kinds of materials and sorted out simple and easy-to-use methods of operation. I hope it will be helpful to answer the doubts of "what is real-time migration of docker containers?" Next, please follow the editor to study!

What exactly is live migration?

Container real-time migration refers to the process of moving applications between different physical computers or cloud platforms without interruption of client access. The memory, file system, and network connections needed to run the container on bare metal hardware can be transferred from the source host to the target host, while ensuring that the state is continuous.

Problems that can be solved by live migration

Live migration technology can solve a variety of problems:

Downtime during hardware maintenance. When the system administrator needs to upgrade the hardware, the process of moving all customers from one hardware node to another is extremely painful, and in many cases it is not even possible without downtime.

Unbalanced cluster load. When a hardware node begins to overload, the rebalancing process may need to execute specific application patterns to narrow the selection of workloads that can run in the cluster.

Cloud platform failure. At present, there are many cloud platforms on the market, and sometimes these cloud platforms will shut down, change the pricing strategy, or deteriorate the quality of service. In most cases, how to easily migrate applications from one cloud vendor to another is a big problem.

Alternative solution

The above problems can be solved in many ways. Let's take a look at how people solve these problems in ways other than live migration.

Planned downtime. In order to maintain the cluster, the owner of the application can announce the maintenance window and possible outages in advance, then shut down the hardware and reboot after all changes have been completed. The problem with this approach is that downtime can be very long.

Traffic is rerouted. For maintenance purposes, you can restore a copy of each application to another hardware node, then reroute traffic to the new copy and close the original copy. The problem with this approach is complexity: applications must be explicitly designed for this situation to achieve high availability and data synchronization. In addition, this approach requires more hardware resources.

Micro services. The application service is further subdivided into multiple containers and the containers are dispersed on different physical servers to avoid downtime caused by hardware failures. Affected containers can be automatically restored on active hardware nodes. However, the problem with this approach is still complex, because applications in the cluster must be properly designed to achieve high availability and restore after failure.

How Live Migration works

Use the following structure to see how the live migration process is technically implemented.

Source node-location of containers before real-time migration

Target node-location of container after real-time migration

In order to perform a live migration, the platform first needs to freeze the container of the source node, block memory, processes, file systems, and network connections, and obtain the status of the container. The content is then copied to the target node. The platform restores the state on the target node, thaws the container, and performs a quick cleanup process on the source node.

The whole process is quite intuitive: get the state, copy the state, restore the state. Note, however, that the container will be frozen for a period of time, so this issue needs to be taken into account when designing the application architecture, which can cause problems for some applications.

There are two types of live migration solutions. One is pre-replication memory (Pre-copy memory). If the container is to be migrated, the platform starts tracking the memory usage of the source node and copies the memory to the target node in parallel until the memory difference between the two nodes is minimized. Then the platform freezes the container, acquires the remaining status data, migrates it to the target node, and restores and thaws the container.

Another solution is replicated memory (Post-copy memory), also known as lazy migration (Lazy migration). At the beginning, the system freezes the container in the source node, gets the state from the rapidly changing memory page, moves the state to the target node and restores it, and then thaws the container. The remaining state information is then copied from the source node to the target node in background mode.

Depending on the application, each container usually needs to be frozen for 5-30 seconds. This time is very short compared to the hours of downtime during cluster maintenance.

Use cases for real-time migration

Non-stop hardware maintenance

In the maintenance window, containers can be migrated from one physical hardware node to another in a unified data center in real-time mode without causing downtime at all.

Load balancing

Achieve load balancing by migrating containers from one hardware node to another through real-time migration. All this can even be done automatically by implementing the corresponding scheduling algorithms and triggers.

High availability within the same hardware area and data center

Cloud service providers can preconfigure and provide a range of hardware availability areas in one or more data centers, so end users can perform real-time migration of containers without the intervention of a system administrator. As a result, more high availability options are available.

Change the cloud provider

Live migration can also help end users avoid locking up by cloud infrastructure vendors. Applications can be migrated to another cloud service provider without any reconfiguration or redeployment operations.

Bottlenecks and potential problems

If you want to solve the above problems through a live migration solution, you also need to consider the various challenges you may face:

During real-time migration, performance degradation may occur after the container is frozen. This is a big problem for some applications that may not be able to accept any performance degradation, such as high-load monolithic real-time applications. But for most applications on the Internet, especially Web applications, a short-term freeze is acceptable.

Another challenge is related to big data and data that cannot be easily moved between different cloud providers because of rapid change. Network latency and data volume may be the biggest obstacles to successful real-time migration.

Public IP in a cloudy environment. Containers that use public IP addresses cannot be migrated between different cloud providers because IP addresses can only be left in each vendor's own network.

If applications in the container use native API or native cloud services from specific cloud service providers, real-time migration across the cloud may be difficult or even impossible.

Real-time migration products on the market

Which companies currently offer live migration products? There are already several products that can migrate containers in real time.

Virtuozzo-this is the team that developed live migration technology for containers. They are pioneers in this field and have provided container engines that can be used in production environments, including real-time migration capabilities.

RunC, derived from Open Containers Initiative, is another container engine solution that will include live migration capabilities based on CRIU.

The container orchestration platform provided by Jelastic provides production applications with the ability to migrate in real time across hardware areas, data centers, and cloud providers.

The real-time migration of containers is still a relatively new technology. For enterprises, however, the benefits of this technology are obvious: no downtime maintenance, no need to spend a lot of time preparing, confirming, and reconfirming everything. So this solution is a good way to improve availability and gain more flexibility. If you have your own experience in moving containers from one instance across the data center to another in real-time mode, welcome to share with us.

At this point, the study on "what is the real-time migration of docker containers" is over. I hope to be able to solve your doubts. The collocation of theory and practice can better help you learn, go and try it! If you want to continue to learn more related knowledge, please continue to follow the website, the editor will continue to work hard to bring you more practical articles!

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.