A generic architecture for autonomic service and network management☆
Introduction
Internet protocol (IP) networks have been growing dramatically in size and functionality in the past decade, and are evolving into a global service communication infrastructure. In addition to the traditional best-effort data services, quality of service (QoS) guaranteed telecommunication services have started to be deployed over IP networks, for example, Voice over IP (VoIP). To reduce the time-to-market of new Internet services and lessen the operation/development/capital costs of service providers (SPs), it is necessary to develop a new service delivery framework by which the SPs can create and deploy QoS guaranteed services over IP in a scalable, flexible, and automatic way. From the information technology (IT) world, autonomic computing [1], [2] is touted as the means to providing a rich set of IT services over a common computing infrastructure. The key feature of autonomic computing is the automated management of computing resources, encompassing the characteristics of self-configuration, self-optimization, self-healing, and self-protection. The application of autonomic management principles to ensure the delivery of telecommunications services over IP networks is largely unexplored. In this paper, we introduce an Autonomic Service Architecture (ASA) to address this need.
Many of the studies on autonomic computing or autonomic management [3], [4], [5], [6] focus on the application of the autonomic concept to a certain service or application environment. In this paper, we propose a generic architecture, ASA, for autonomic service delivery over IP networks. ASA is driven by our view that “every thing is a service”, from complex multimedia applications to simple IP packet delivery, and all the services are organized into a service hierarchy. Using this service perspective, ASA provides a uniform framework for service and transport network management. The underlying IP packet delivery and queueing are considered as basic services, upon which upper layer applications are built as composite services. We will show that ASA enables the network to orchestrate by itself the service, resource, billing, and fault management under the high-level policy guidance, where interaction with the human network managers is limited to specifying the services according to customer needs and establishing the management policies according to the QoS and revenue objectives.
To illustrate ASA’s operation, we apply it to manage a multiprotocol label switching (MPLS) [7] based differentiated services (DiffServ) [8] network. DiffServ/MPLS is a promising IP network infrastructure due to its scalable QoS management and its traffic engineering capability [9]. We will demonstrate the VoIP service delivery through a virtual network (VN) [10] over the DiffServ/MPLS transport network, which is managed following the hierarchical service composition used by ASA. A representative framework for resource management and traffic engineering in DiffServ/MPLS networks is the TEQUILA architecture proposed in [11]. In the TEQUILA architecture, a DiffServ/MPLS network is operated in a “first plan, then take care” fashion, first through off-line planning and dimensioning and subsequently through dynamic operations and management functions for self-optimization. It will be shown that the TEQUILA architecture maps to the generic ASA framework, behaving as an instance of ASA for the management of DiffServ/MPLS IP transport networks.
Although ASA supports autonomic management by proposing a generic architecture, this architecture needs to be solidified and materialized through specific technologies, such as service level agreement (SLA) negotiation, policy control, efficient resource allocation, as well as automatic account and billing management. Note that SLAs are contracts between SPs and customers that define, among others, the services provided, the metrics associated with these services, acceptable and unacceptable service levels, liabilities on the part of the SP and the customer, and actions to be taken in specific circumstances. SLAs are critical in guaranteeing service delivery. Service management must ensure that necessary resources are provided to meet the SLA.
In this paper, we also investigate implementation details for autonomic service and resource management in the DiffServ/MPLS network. We propose an MPLS label stacking technique and path-oriented bandwidth management to support VN-based service provisioning. Particularly, we focus on the efficient resource management, which is a critical problem for all SPs desiring higher revenue. ASA is a SLA-centric management model, where it can be considered that the resources are shared by all SLAs over the network. We develop an autonomic bandwidth borrowing scheme for efficient inter-SLA resource sharing in a DiffServ/MPLS network. With bandwidth borrowing, the network can automatically adjust the resource allocation to each SLA when the traffic load conditions deviate from the engineered operation point or the high level policies change, so that the spare capacity in underloaded SLAs1 can be exploited and QoS specification of all the SLAs are always guaranteed.
The remainder of this paper is organized as follows. In Section 2, we give a review of related work. Section 3 describes ASA. Section 4 illustrates the operation of the autonomic resource broker (ARB), the key component of ASA. Section 5 shows how ASA is applied to the management and control of a DiffServ/MPLS network. Section 6 presents the bandwidth borrowing scheme. Section 7 presents the computer simulation results. Section 8 gives the concluding remarks.
Section snippets
Autonomic computing in IT services
The IBM Autonomic Computing Architecture [1], [2] is the pioneer work in the new wave of autonomics, which defines an abstract information framework for self-managing IT systems. In the information framework, an autonomic system is a collection of autonomic elements. Each autonomic element consists of an autonomic manager (AM) and the managed resource (MR). The communication between the AM and the MR is done through the MR’s management interfaces, which exposes two types of hooks, sensors and
Autonomic service architecture
In this section, we will present ASA according to a layered view, where services are built on top of virtual and physical resources. The players involved in the delivery of a service are the customers and the SPs. After customers and SPs negotiate the services needed and their corresponding SLAs, ASA will manage these services in order to ensure satisfactory service delivery without SP’s intervention. It is noteworthy that some manual actions will still be needed to complete the service
Autonomic resource broker architecture
Autonomic resource brokers are the autonomic components, which constitute the autonomic service architecture. Fig. 3 shows the ARB’s internal architecture.
Service example: VoIP over DiffServ/MPLS
In this section, we illustrate ASA’s operation by applying it for autonomic management of a SIP-based VoIP service that is delivered over a DiffServ/MPLS IP transport network. We consider that the IP transport network supports multiple types of upper-layer services; one of the services is VoIP. According to ASA, the VoIP service is a composite service consisting of the following virtual resources: SIP User Agents (UAs) at the customer premises, SIP Signaling Servers (proxy, redirect,
Autonomic inter-SLA resource sharing
The autonomic service architecture is a SLA-centric management system. At the SLA level, the transport network resources are shared by a set of SLAs. For each SLA, the resource requirement is determined in accordance with the management policy to guarantee QoS under an engineered traffic load (which is the estimated long-term average traffic demand). The planning component of the core network CARB will find an optimal solution to accommodate all the SLA resource requirements by the network
Performance evaluation
In this section, we use a case study to demonstrate the performance of the bandwidth borrowing scheme. The units used for related measures are second for time, capacity unit (c-unit) for link/trunk/SLA capacity and efficient bandwidth usage, call/second for call arrival rate, and c-unit/call for effective bandwidth. The network topology, SLA and trunk deployment for the case study are shown in Fig. 7, where 8 edge routers and 1 core router are connected through 12 links. Five SLAs are supported
Concluding remarks
This paper presents a general approach to autonomic service management using ASA, which will allow service providers to reduce the costs of delivering services to customers, and to manage services and network resources under a uniform framework. ASA is based on two main concepts: virtualization of physical resources using a Common Resource Format, and autonomic service delivery using a hierarchy of Autonomic Resource Brokers. The CRF allows the service delivery framework applicable to all types
Yu Cheng received the B.E. and M.E. degrees in Electrical Engineering from Tsinghua University, Beijing, China, in 1995 and 1998, respectively, and the Ph.D. degree in Electrical and Computer Engineering from the University of Waterloo, Waterloo, ON, Canada, in 2003.
From September 2003 to August 2004 he was a postdoctoral fellow in the Department of Electrical and Computer Engineering, University of Waterloo. From September 2004 to August 2006, he was a postdoctoral fellow in the Department of
References (38)
- et al.
The vision of autonomic computing
IEEE Computer
(2003) - IBM Corporation, An architectural blueprint for autonomic computing, White Paper...
Autonomia: an autonomic computing environment
Proceedings of the IEEE International Performance, Computing, and Communications Conference
(2003)AutoMate: enabling autonomic applications on the grid
Proceedings of the Autonomic Computing Workshop
(2003)- IBM and University of Berkeley, Oceano Project,...
- et al.
Adaptive control overlay for service management
Proceedings of the Workshop on the Design of Self-Managing Systems, International Conference on Dependable Systems and Networks (DSN)
(2003) - et al.
Multiprotocol label switching architecture
IETF RFC 3031
(2001) - et al.
An architecture for differentiated services
Internet RFC 2475
(1998) Multi-protocol label switching (MPLS) support of differentiated services
IETF RFC 3270
(2002)- et al.
Virtual network resource management for next-generation networks
IEEE Commun. Mag.
(2003)
A management and control architecture for providing IP differentiated services in MPLS-based networks
IEEE Commun. Mag.
Mobile service agents enabling intelligence on demand in telecommunications
Proceedings of IEEE GLOBECOM’96
Distributed management with mobile components
Proceedings of the 6th IFIP/IEEE International Symposium on Integrated Network Management
Accord: a programming framework for autonomic applications
IEEE Trans. Syst., Man. Cybern. C.
The open grid services architecture: where the grid meets the web
IEEE Internet Comput.
Oceano: SLA based management of a computing utility
Proceedings of the IEEE/IFIP International Symposium on Integrated Network Management
Self-aware management of IP networks with QoS guarantees
Int. J. Network Mgmt.
Cited by (33)
DDP: A Dynamic Dimensioning and Partitioning model of Virtual Private Networks resources
2012, Computer CommunicationsCitation Excerpt :Having the efficient routing schemes, VPNPs can allocate the incoming traffic over the optimal routes. If the VPNP provider is using a label-switched architecture (e.g. MPLS [6]), this can be achieved by mapping the calculated routes to forwarding labels. It is clear from the table that the larger the number of requests is the longer the simulation CPU time is.
A distributed resource management model for Virtual Private Networks: Tit-for-Tat strategies
2012, Computer NetworksCitation Excerpt :However, in many cases, the VPN operators may have some inputs to predict their short-term future resource requirements. For example, in the Voice over IP (VoIP) video conferencing application [18], the VPN operator can monitor the Session Initiation Protocol (SIP) messages [32] which negotiate the parameters of multimedia session [6]. Therefore, it should be better to rely on the operators to forecast their resource requirement.
A collusion-resistant mechanism for autonomic resource management in Virtual Private Networks
2010, Computer CommunicationsCitation Excerpt :Consequently, the need is emerging to find an alternative resource management models that overcome the current limitations. Autonomic Service Architecture (ASA) is proposed in [4] to cope with the above limitations by creating a uniform framework for automated management. ASA ensures services’ delivery based on a Service level Agreements (SLAs) that have been conducted between customers and ISPs.
ABLA: Application-Based Load-Balanced Approach for Adaptive Mapping of Datacenter Networks
2023, Electronics (Switzerland)A fair model for autonomic bandwidth management in virtual private networks
2016, 4th IEEE Conference on Communication, Networks and Satellite, COMNESTAT 2015 - ProceedingsTowards autonomic cloud configuration and deployment environments
2015, Proceedings - 2014 International Conference on Cloud and Autonomic Computing, ICCAC 2014
Yu Cheng received the B.E. and M.E. degrees in Electrical Engineering from Tsinghua University, Beijing, China, in 1995 and 1998, respectively, and the Ph.D. degree in Electrical and Computer Engineering from the University of Waterloo, Waterloo, ON, Canada, in 2003.
From September 2003 to August 2004 he was a postdoctoral fellow in the Department of Electrical and Computer Engineering, University of Waterloo. From September 2004 to August 2006, he was a postdoctoral fellow in the Department of Electrical and Computer Engineering, University of Toronto, ON, Canada. He joins the Department of Electrical and Computer Engineering, Illinois Institute of Technology, Chicago, USA as an Assistant Professor in September 2006. His research interests include service-oriented networking, autonomic network management, Internet performance analysis, quality of service provisioning, wireless networks, and wireless/wireline interworking. He received a Postdoctoral Fellowship Award from the Natural Sciences and Engineering Research Council of Canada (NSERC) in 2004. He is a Member of IEEE and ACM.
Ramy Farha is a Ph.D. candidate in the Department of Electrical and Computer Engineering, University of Toronto, Toronto, Ontario, Canada. He received his B.Eng degree from the American University of Beirut in 2001, and his M.A.Sc degree from the University of Toronto in 2003. He received the Natural Sciences and Engineering Research Council (NSERC) PGS award and the Ontario Graduate Scholarship (OGS) award, both in 2003. In addition, he holds the Distinguished Graduate Award from the American University of Beirut in 2001. His research interests include IP mobility, passive optical networks, peer-to-peer and overlay networks, and autonomic service management.
Myung Sup Kim received the BS, MS and PhD degrees in computer science and engineering from Pohang University of Science and Technology (POSTECH), Korea, in 1998, 2000 and 2004, respectively. From September 2004 to July 2006, he was a post-doctoral fellow in the Department of Electrical and Computer Engineering, University of Toronto, Canada. He joins the Department of Computer and Information Science, Korea University, Jochiwon, Korea, as an Assistant Professor in August 2006. His research interests include service and network management, Internet traffic monitoring and analysis, application-level networking, network-security attack detection and prevention, and autonomic network resource management. He is a member of KNOM.
Alberto Leon-Garcia received the B.S., M.S., and Ph.D. degrees in electrical engineering from the University of Southern California, in 1973, 1974, and 1976 respectively. He is a Full Professor in the Department of Electrical and Computer Engineering, University of Toronto, ON, Canada, and he currently holds the Nortel Institute Chair in Network Architecture and Services. In 1999 he became an IEEE fellow for “For contributions to multiplexing and switching of integrated services traffic”.
Dr. Leon-Garcia was Editor for Voice/Data Networks for the IEEE Transactions on Communications from 1983 to 1988 and Editor for the IEEE Information Theory Newsletter from 1982 to 1984. He was Guest Editor of the September 1986 Special Issue on Performance Evaluation of Communications Networks of the IEEE Selected Areas on Communications. He is also author of the textbooks Probability and Random Processes for Electrical Engineering (Reading, MA: Addison-Wesley), and Communication Networks: Fundamental Concepts and Key Architectures (McGraw-Hill), co-authored with Dr. Indra Widjaja.
James Won-Ki Hong is an associate professor in the Dept. of Computer Science and Engineering, POSTECH, Pohang, Korea. He received a Ph.D. degree from the University of Waterloo, Canada in 1991 and an M.S. degree from the University of Western Ontario in 1985. He has worked on various research projects on network and systems management, with a special interest in Web, Java, CORBA, and XML technologies. His research interests include network and systems management, distributed computing, and network monitoring and planning. He has served as Technical Chair (1998–2000), Vice Chair (2003–2005) and Chair (2005-present) for IEEE Comsoc CNOM. He is also serving as Director of Online Content for the IEEE Comsoc (Jan. 2004-Dec. 2005). He is a NOMS/IM Steering Committee Member and a Standing Committee Member of APNOMS. He was technical co-chair of NOMS 2000 and APNOMS ’99. He was Finance Chair for IM 2005 and Finance Chair and Chair of Local Planning Committee for NOMS 2004. He is an editorial advisory board member of JNSM and IJNM. He is also editor-in-chief of KNOM Review Journal. He is a member of IEEE, KICS, KNOM, and KISS.
- ☆
This work is supported by a Postdoctoral Fellowship and a Discovery Grant from Natural Sciences and Engineering Research Council of Canada (NSERC).