Programmable real-time computing and network load awareness method for decentralized computing network

Abstract

The invention discloses a programmable real-time computing and network load sensing method for a distributed computing network, which comprises the following steps of: (1) realizing the computing of anode, the network load and the state collection in a switch chip through a single-node load and state sniffing technology; synchronizing the information to other decentralized computing nodes throughthe reconstructed protocol message; (2) realizing load and state information collection of each node in a network flow transmission process in the switch chip through an end-to-end load and state sniffing technology, carrying the information in the network flow in a stack manner, and collecting the path information at a tail node; and (3) setting a flow acquisition point through calculation, network state acquisition and threshold calculation, thereby realizing analysis of all flows in the monitored network. According to the invention, the real-time load, time delay and jitter perception of the network is realized, and the capability requirements of a decentralized computing network on programmable perception and programmable scheduling are met.

Description

technical field [0001] The invention relates to the technical field of computer networks, in particular to a programmable real-time computing and network load sensing method for decentralized computing networks. Background technique [0002] In a distributed computing network, the network awareness capability provides the task scheduling system with computing node information collection and maintenance functions, which is a key link in a decentralized computing network. Currently in foreign countries, network perception technology is in the research stage, and a batch of achievements on network status perception in decentralized computing networks have been formed. At the same time, a test platform has been built, and theoretical verification has been carried out through IPERF, PING, and OWAMP to analyze the network under different loads. Packet loss, time delay and jitter data under the circumstances, foreign research in this field has not yet entered the industrialization ...

AI Technical Summary

Problems solved by technology

[0006] In short, the main problems existing in the existing network perception technology: the current technology has not been able to go deep into the hardware of the network equipment, and has the real-time load, delay and jitter perception of the network

Technically, it cannot meet the requirements of numerous distributed computing nodes, rapid changes in network status in real time, and quantitative analysis of network load by computing tasks.

And the technological evolution in the era of SDN has not yet met the capability requirements of the decentralized computing network for programmable perception and programmable scheduling

Images