Skip to main content
eScholarship
Open Access Publications from the University of California

Enhancing Automated Network Management

  • Author(s): Wang, Huazhe
  • Advisor(s): Qian, Chen
  • et al.
Abstract

Network management benefits from automated tools. With the recent advent of software-defined principles, automated tools have been proposed from both industry and academia to fulfill function components in the network management control loop. While automation aims to accommodate the ever increasing network diversity and dynamics with improved reliability and management efficiency, it also brings new concerns as it’s becoming more difficult to understand the control of the network and operators cannot rely on traditional troubleshooting tools. Meanwhile, how to effectively integrate new automation tools with existing legacy networks remains a question. This dissertation

presents efficient methods to address key functionalities within the control loop in the adaption of automated network management.

Identifying the network-wide forwarding behaviors of a packet is essential for many network management tasks, including policy enforcement, rule verification, and fault localization. We start by presenting AP Classifier. AP Classifier was developed based on the concept of atomic predicates which can be used to characterize the forwarding behaviors of packets. There is an increasing trend that enterprises outsource their Network Function (NF) processing to a cloud to lower cost and ease management. To avoid threats to the enterprise’s private information, we propose SICS based on AP Classifier, a secure and dynamic NF outsourcing framework. Stateful NFs have become essential parts of modern networks, increasing the complexity in network management. A major step in network automation is to automatically translate high level network intents into low level configurations. To ensure those configurations and the states generated by automation match intents, we present Epinoia, a network intent checker for stateful networks. While the concept of auto-translation sounds promising, operators may not know what intents should be. To close the control loop, we present AutoInfer to automatically infer intents of running networks, which helps operators understand the network runtime states.

Main Content
Current View