What Is Compute-Network Collaboration?

Why Do We Need Compute-Network Collaboration?

On an intelligent lossless network, there are a large number of servers and network devices. To achieve lossless transmission, the servers and network devices must have the same the QoS configurations, such as lossless queues and DSCP. Traditional manual configuration, however, not only involves heavy workload but is also prone to errors. To address this issue, the intelligent lossless network provides the compute-network collaboration function, which automatically synchronizes QoS configurations from the network side to the server side. This ensures consistency between server-side and network-side QoS configurations while reducing configuration workload.

How Does Compute-Network Collaboration Work?

Implementation of network-compute collaboration

As shown in the preceding figure:

1. The network-side switch generates the mapping between lossless service types, queues, and DSCP values based on its QoS configuration (including the lossless service types, corresponding DSCP values, and default DiffServ domain configuration), and sends the mapping to the server.
2. After receiving the mapping information, the server synchronizes its configuration with the network-side configuration. When sending lossless service packets, the server sets DSCP values for the packets.