Invited PaperA new traffic engineering manager for DiffServ/MPLS networks: design and implementation on an IP QoS Testbed
Introduction
One of the most actively studied open issues in several areas of communication networks is the problem of bandwidth reservation and management. Load balancing is another important issue. It is desirable to avoid portions of the network becoming over-utilized and congested, while alternate feasible paths remain underutilized. These issues are addressed by Traffic Engineering (TE). The MultiProtocol Label Switching (MPLS) technology is a suitable method to provide TE, independent of the underlying layer2 technology [1], [2]. MPLS per se cannot provide service differentiation, which brings up the need to complement it with another technology capable of providing such feature: DiffServ. DiffServ is becoming prominent in providing scalable network designs supporting multiple classes of services. When optimization of resources is sought, DiffServ mechanisms need to be complemented by existing MPLS traffic engineering mechanisms, which then becomes DiffServ-aware Traffic Engineering (DS-TE) [3], currently under discussion in Internet Engineering Task Force (IETF). In this case, DiffServ and MPLS both provide their respective benefits. It is obvious that such future networks cannot be managed manually when all new protocols are implemented. Therefore, automated managers need to be developed to simplify network management and to engineer traffic efficiently [4].
With the objective to studying and researching the issues mentioned above, we assembled an IP QoS testbed in our laboratory (http://www.ece.gatech.edu/research/labs/bwn). The testbed is a high-speed top-of-the-line mix of highly capable routers and switches for testing DiffServ and MPLS functionalities. During our experiences with the testbed, we realized the need for an improved set of algorithms for network management and also an integrated architecture for an automated network manager. This led to the design and implementation of Traffic Engineering Automated Manager (TEAM).
Individual problems addressed by TEAM may already have been considered, but to the best of our knowledge, an integrated solution does not exist. We are developing TEAM as a centralized authority for managing a DiffServ/MPLS domain. Our proposal is a comprehensive study that describes practical solutions for MPLS network management. TEAM is responsible for dynamic bandwidth and route management. Based on the network state, TEAM takes the appropriate decisions and reconfigures the network accordingly. TEAM is designed to provide a novel and unique architecture capable of managing large scale MPLS/DiffServ networks.
The structure of the rest of the paper is as follows. In Section 2, we enlist the components of our IP QoS testbed. The following section, Section 3, includes a design description of TEAM along with comparison with other MPLS network management tools. In Section 4 we present our proposed algorithms for bandwidth management, namely Label Switched Path (LSP) setup and dimensioning, LSP preemption and LSP capacity allocation. Section 5 discusses the route management aspects of TEAM, followed by the description of the measurement tool employed by TEAM in Section 6. In Section 7, we present the implementation details for TEAM. Finally, we conclude the paper in Section 8.
Section snippets
Physical testbed
We have a full-fledged Next Generation Internet routers physical testbed in our Broadband and Wireless Networking Laboratory (BWN-Lab) at Georgia Institute of Technology, equipped with DiffServ capable routers and switches manufactured by Cisco. We have a Cisco 7500 router with a Gigabit Ethernet card and a layer 3 switch Catalyst 6500 with an enhanced Gigabit Ethernet card and also other routers and switches. These routers and switches are widely deployed in the backbones of current high-speed
Team Traffic Engineering automated Manager
The design and management of an MPLS network is a fundamental key to the success of the QoS provisioning. Many problems need to be solved such as LSP dimensioning, set-up/tear-down procedures, routing, adaptation to actual carried traffic, preemption, initial definition of the network topology, etc. To illustrate the inter-relations of the listed problems for MPLS network management, let us consider the scenario where network planning methods have provided an initial topology of the MPLS
Bandwidth management
Bandwidth management deals with managing the resources of an MPLS network in an efficient manner to meet QoS requirements. It comprises of LSP setup and dimensioning (Section 4.1), preemption (Section 4.2), and capacity allocation (Section 4.3). In the event of an LSP setup request, the LSP preemption and capacity allocation functions are invoked. In the case of bandwidth reservation request, LSP setup and dimensioning procedures are triggered which may in turn initiate the LSP creation steps
Route management
Route management deals with deciding the routes for LSPs over a physical network (Section 5.1) and for bandwidth requests over an MPLS network. It is triggered by the arrival of either an LSP setup request or a bandwidth reservation requests in MPLS networks.
Measurement/performance evaluation tool
There are various measurable quantities of interest that can be insightful about the state of the network. Available bandwidth (together with other metrics like latency, loss, etc.) can predict the performance of the network. Based on the bandwidth available, the network operator can obtain information about the congestion in the network, decide the admission control, perform routing, etc. For MPLS networks, the available bandwidth information can be used to decide about the LSP setup [7], LSP
SNMPv3
SNMP is the protocol used for communication between the routers and our manager. Despite the recent vulnerabilities, SNMP is widely accepted as the de facto standard for network management and patches are being released to fix it.
A major deficiency of the current version of SNMP is information security. Community strings are transported as clear text and if compromised, an attacker could have access to all management information and configuration rights. Preventive measurements such as packet
Conclusions
We propose the design of a set of new algorithms to provide QoS and better resource utilization in an MPLS network and an integrated architecture, TEAM, as an MPLS domain manager. The new algorithms concern resource management and route management. All algorithms will be developed and evaluated through simulation and experiments individually on our physical testbed. Now, we are focusing on their inter-working and the development of TEAM as a whole. We have a full-fledged physical testbed, with
Acknowledgements
This work was supported by NASA Goddard. The work of J. C. de Oliveira was also supported in part by CAPES (The Brazilian Ministery of Education Agency).
References (61)
- et al.
Decentralized network connection preemption algorithms
Computer Networks and ISDN Systems
(1998) - D.O. Awduche, J. Malcom, J. Agogbua, M. O'Dell, J. McManus, Requirements for Traffic Engineering over MPLS, IETF RFC...
- D.O. Awduche, L. Berger, D. Gan, T. Li, V. Srinivasan, G. Swallow, RSVP-TE: Extensions to RSVP for LSP Tunnels, IETF...
- D. Awduche, A. Chiu, A, Elwalid, I. Widjaja, X. Xiao, Overview and Principles of Internet Traffic Engineering, IETF RFC...
Predictions for the core of the network
IEEE Internet Computing
(2000)- et al.
Proceedings of IEEE International Conference on Telecommunications, Bucharest, Romania
(2001) - et al.
Proceedings of 17th International Teletraffic Congress'01, Salvador, Brazil
(2001) - et al.
Optimal policy for LSP setup in MPLS networks
Computer Networks
(2002) - et al.
Proceedings of IEEE INFOCOM'02, New York, USA
(2002) - T. Anjali, C. Scoglio, I.F. Akyildiz, G. Uhi, A New Scheme for Traffic Estimation and Resource Allocation for Bandwidth...
RATES: a server for MPLS traffic engineering
IEEE Network
Minimum interference routing of bandwidth guaranteed tunnels with MPLS traffic engineering application
IEEE Journal on Selected Areas in Communications
Proceedings of IEEE INFOCOM'01, Anchorage, USA
Proceedings of IEEE International Conference on Telecommunications, Bucharest, Romania
Traffic engineering with MPLS in the Internet
IEEE Network Magazine
Proceedings of Quality of Future Internet Services'00, Berlin, Germany
Proceedings of IEEE GLOBECOM'01, San Antonio, USA
Proceedings of IEEE GLOBECOM'00, San Francisco, USA
Eighth International Workshop on Quality of Service
Markov Decision Processes: Discrete Stochaftic Dynamic Programming
Proceedings of MILCOM'98
Proceedings of IEEE ICC'98
Proceedings of IEEE INFOCOM'92
Quality of Service in IP Networks
Traffic theory the Internet
IEEE Communications Magazine
An optimal bandwidth allocation strategy for the delivery of compressed prerecorded video
ACM/Springer Verlag Multimedia Systems Journal
Proceedings of IEEE INFOCOM'96, San Francisco, USA
Cited by (20)
A roadmap for traffic engineering in software defined networks
2014, Computer NetworksDesign of priority-based active queue management for a high-performance IP switch
2013, Computers and Electrical EngineeringCitation Excerpt :To ensure high bandwidth for high-speed multimedia applications with suitable QoS requirements over an IP switch, Active Queue Management (AQM) techniques are needed. AQM includes both buffer management and packet scheduling mechanisms, which are closely related and affect almost all QoS metrics [1]. For example, a queue is more effectively served by using a scheduling technique, which implies shorter delays, lower drop rates, and higher throughput.
Inter-autonomous system provisioning for end-to-end bandwidth guarantees
2007, Computer CommunicationsInternet like control for MPLS based traffic engineering: Performance evaluation
2005, Performance EvaluationReview on traffic engineering and load balancing techniques in software defined networking
2021, Lecture Notes in Networks and Systems