A Stateful Data Plane Fault Detection Approach Using Policy Decomposition and Symbolic Execution

Abstract

The invention discloses a stateful data plane fault detection method adopting policy decomposition and symbolic execution. By constructing a network function model and decomposing context-related strategies, a symbolic execution engine is used for symbolic execution to generate an original test case; and then the original test case is translated into Legal test data flow that conforms to the data flow specification; inject the test data flow into the data plane; compare the actual test data flow processing results with the test data flow theoretical processing results, so as to realize the fault detection of the stateful data plane and model the internal logic of the network function , can verify the internal logic of the intermediate network equipment, and can handle runtime failures of the data plane due to software errors, hardware failures, etc.; use test data streams for detection, can directly detect the state of the data plane, and has good scalability . Utilizing policy decomposition and symbolic execution reduces the size of the state space that needs to be explored and improves the scalability of fault detection.

Description

technical field [0001] The invention relates to the field of network technology, in particular to a stateful data plane fault detection method using policy decomposition and symbolic execution. Background technique [0002] The Internet has changed people's production and lifestyle since the industrial age, changed the traditional economic model, and driven the progress of world civilization. With the continuous emergence of emerging technologies such as cloud computing, Internet of Things, blockchain, and big data, the network carries more and more functions. Today, the network is more than just a means of communication connecting various computing devices, it has become one of the most important critical infrastructures of modern society. [0003] Network failure is a problem that network administrators must face. A survey for network administrators shows that 80% of network administrators manage networks with at least 10 faults per month on average, and 35% of network a...

AI Technical Summary

Problems solved by technology

[0009] The main problem of the above scheme 1 is that the modelers need to learn and master the SEFL language, and use the SEFL language to model the intermediate network devices existing in the network, which has poor versatility and will cause high network maintenance costs

Since the network configuration is verified on the control plane, it can only detect configuration errors in the control plane, but cannot handle runtime failures in the data plane due to software errors, hardware failures, etc.

[0010] The main problem of the above scheme 2 is: because it gives up the verification of the internal processing logic of the intermediate network equipment, its application scope and practicability are greatly reduced, and it is far from meeting the actual network fault detection requirements

Similarly, VMN only verifies the network configuration on the control plane, and cannot verify whether the actual forwarding and processing behavior of the data plane is consistent with that of the control plane. This also limits its detection capabilities and cannot directly detect the status of the data plane.

[0011] The main problem of the above scheme 3 is: because it needs to test the entire data plane as a whole, the scalability is very poor. With the increase of the network scale and the complexity of the network strategy, its time overhead increases exponentially, which limits Its scope of application and practicality

Images