Practical Network Management Research Articles

Traffic Refinery

Network management often relies on machine learning to make predictions about performance and security from network traffic. Often, the representation of the traffic is as important as the choice of the model. The features that the model relies on, and the representation of those features, ultimately determine model accuracy, as well as where and whether the model can be deployed in practice. Thus, the design and evaluation of these models ultimately requires understanding not only model accuracy but also the systems costs associated with deploying the model in an operational network. Towards this goal, this paper develops a new framework and system that enables a joint evaluation of both the conventional notions of machine learning performance (model accuracy) and the systems-level costs of different representations of network traffic. We highlight these two dimensions for two practical network management tasks, video streaming quality inference and malware detection, to demonstrate the importance of exploring different representations to find the appropriate operating point. We demonstrate the benefit of exploring a range of representations of network traffic and present Traffic Refinery, a proof-of-concept implementation that both monitors network traffic at 10~Gbps and transforms traffic in real time to produce a variety of feature representations for machine learning. Traffic Refinery both highlights this design space and makes it possible to explore different representations for learning, balancing systems costs related to feature extraction and model training against model accuracy.

Traffic Refinery

Network management often relies on machine learning to make predictions about performance and security from network traffic. Often, the representation of the traffic is as important as the choice of the model. The features that the model relies on, and the representation of those features, ultimately determine model accuracy, as well as where and whether the model can be deployed in practice. Thus, the design and evaluation of these models ultimately requires understanding not only model accuracy but also the systems costs associated with deploying the model in an operational network. Towards this goal, this paper develops a new framework and system that enables a joint evaluation of both the conventional notions of machine learning performance (e.g., model accuracy) and the systems-level costs of different representations of network traffic. We highlight these two dimensions for two practical network management tasks, video streaming quality inference and malware detection, to demonstrate the importance of exploring different representations to find the appropriate operating point. We demonstrate the benefit of exploring a range of representations of network traffic and present Traffic Refinery, a proof-of-concept implementation that both monitors network traffic at 10~Gbps and transforms traffic in real time to produce a variety of feature representations for machine learning. Traffic Refinery both highlights this design space and makes it possible to explore different representations for learning, balancing systems costs related to feature extraction and model training against model accuracy.

From model to implementation: a network algorithm programming language

Software-defined networking (SDN) is a revolutionary technology that facilitates network management and enables programmatically efficient network configuration, thereby improving network performance and flexibility. However, as the application programming interfaces (APIs) of SDN are low-level or functionality-restricted, SDN programmers cannot easily keep pace with the ever-changing devices, topologies, and demands of SDN. By deriving motivation from industry practice, we define a novel network algorithm programming language (NAPL) that enhances the SDN framework with a rapid programming flow from topology-based network models to C+ implementations, thus bridging the gap between the limited capability of existing SDN APIs and the reality of practical network management. In contrast to several state-of-the-art languages, NAPL provides a range of critical high-level network programming features:linebreak (1) topology-based network modeling and visualization; (2) fast abstraction and expansion of network devices and constraints; (3) a declarative paradigm for the fast design of forwarding policies; (4) a built-in library for complex algorithm implementation; (5) full compatibility with C+ programming; and (6) user-friendly debugging support when compiling NAPL into highly readable C+ codes. The expressiveness and performance of NAPL are demonstrated in various industrial scenarios originating from practical network management.

Promoting Cooperation in Wireless Relay Networks Through Stackelberg Dynamic Scheduling

This paper discusses a new perspective for the application of game theory to wireless relay networks, namely, how to employ it not only as an analytical evaluation instrument, but also in constructively deriving practical network management policies. We focus on the problem of medium sharing in wireless networks, which is often seen as a case where game theory just proves the inefficiency of distributed access, without proposing any remedy. Instead, we show how, by properly modeling the agents involved in such a scenario, and enabling simple but effective incentives towards cooperation for the users, we obtain a resource allocation scheme which is meaningful from both perspectives of game theory and network engineering. Such a result is achieved by introducing throughput redistribution as a way to transfer utilities, which enables cooperation among the users. Finally, a Stackelberg formulation is proposed, involving the network access point as a further player. Our approach is also able to take into account power consumption of the terminals, still without treating it as an insurmountable hurdle to cooperation, and at the same time to drive the network allocation towards an efficient cooperation level.

A practical network management system based on an organization hierarchy mobile agent model

AbstractA mobile communication network is composed of a tremendous number of diverse network elements (NEs), which are distributed over a wide geographical area. The system that monitors messages (including warnings) from the NEs and controls the NEs is called the network management system (NMS). Since the NMS is required to handle a large number of messages, it has been implemented with expensive large‐scale UNIX servers. Thus, the communication carrier is forced to invest a large amount in the hardware and middleware of the NMS. Another problem is that the carrier must follow the vendor strategy, such as migration to next‐generation hardware and OS upgrades. The authors have devised the organization hierarchy agent model (OHAM), in which the external specification of the NMS is represented at the data level as agent behavior, making it possible to add new external specifications quickly and flexibly to the NMS as functions. This paper describes the organization hierarchy mobile agent model (OHMAM), which achieves high processing capacity by upgrading the OHAM agents to mobile agents that move rapidly between small‐scale IA servers. A practical commercial system was formed by using the proposed model, and the results of its basic performance evaluation and reliability evaluation are presented in this paper. In the implementation of the proposed model, IA servers, which are inexpensive and readily available, are used as hardware. Open source software (OSS) such as Linux/JavaVM/PostgreSQL is actively used as OS/middleware. Thus, the total cost of ownership (TCO) of the NMS can be greatly reduced. © 2007 Wiley Periodicals, Inc. Electron Comm Jpn Pt 2, 90(10): 84–102, 2007; Published online in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/ecjb.20391

그리드 네트워크를 위한 모니터링 시스템의 설계 및 구현

그리드는 고속 네트워크로 연결된 지리적으로 흩어져 있는 이종의 자원을 연결하여 서로 협업하기 위한 기반구조이다. 그리드 응용을 수행하는데 있어 안정성과 신뢰성이 보장되기 위해서는 네트워크 자원이 관리되어야 한다. GGF의 그리드 모니터링 구조 워킹 그룹(GMAWG)은 대규모 분산 자원을 넓은 지역에 확장 가능하기 위한 효과적인 모니터링 구조를 제안하였다. 이 논문에서는 실제 네트워크 관리에 있어 GMA를 기반으로 하는 그리드 네트워크 모니터링 시스템을 설계하고 구현하였다. 본 시스템으로 네트워크 관리 센터는 그리드 네트워크의 유연하고 확장성있는 관리를 수행할 수 있을 것이다. The Grid is an infrastructure to connect heterogeneous resources that are scattered over areas with high-speed network and to cooperate with each other. To support Grid applications, network resources should be managed, since the network has to be safe and reliable. The Grid Monitoring Architecture Working Group (GMAWG) of the Global Grid Forum (GGF) proposed an effective architecture to be scalable across wide-area networks and encompass a large number of heterogeneous resources. In this paper, we describe the design and implementation of Grid network monitoring system based on the GMA for practical network management. By this system, network operations center can form a management system flexibly and scalably for Grid network.

SNMP, SNMPv2, and RMON: Practical Network Management, Second Edition [New Books

SNMP, SNMPv2, and RMON: Practical Network Management, Second Edition [New Books