The P2P acceleration feature uses the internal bandwidth resources of compute nodes to distribute images among the nodes. This feature can accelerate image pulling to reduce the time to deploy applications. If a large number of nodes in a container cluster need to pull an image at the same time, you can use the P2P acceleration feature to accelerate the pulling.

When a large number of nodes in a container cluster pull an image at the same time, the network bandwidth of the Object Storage Service (OSS) buckets becomes a performance bottleneck. As a result, it takes a long time to pull the image to nodes. The P2P acceleration feature uses the bandwidth resources of your compute nodes to distribute images among the nodes. This relieves the pressure on OSS buckets, improves the speed to pull images, and reduces the time to deploy applications. In a previous test, 1,000 nodes pulled an image of 1 GB in size at the same time in P2P mode and then in regular mode where the bandwidth is 10 Gbit/s. The test result shows that the P2P mode reduces the image pulling time by 95% or more, compared with the regular mode.

You can use the P2P acceleration feature in the following scenarios: