Computer networks are becoming more complex and unmanageable by network engineers effort alone. Management, troubleshooting and configuration of networks have become increasingly complex over time. A means is required to dynamically configure and authenticate the hierarchy in response to changing network topology and available server and network resources. Autonomous network management system can be deployed to automatically manage, Self-configuration, self-healing, self-optimization and self-protection, self-troubleshooting, and restore connectivity without human intervention (Jawahar & Elango (2001). This paper defines autonomous network management, discusses the techniques and goals. A list of the advantages and disadvantages of autonomous network management plus the comparison of autonomous network management compared to the normal methods will also be drawn at the end.
Academics such as Jawahar and Elango (2001) are of the opinion that large-scale networks are becoming unmanageable with human effort alone. Such networks are often widely geographically distributed and consist of many hundreds and thousands of terminal devices. Switching centers can consist of dozens of nodes with numerous different vendor equipments and network protocols. A possible solution could be to enable modern, networked computing systems to manage themselves without direct human intervention.
This paper is of the opinion that the term "Autonomous Network Management" (ANM) has been excessively ascribed to many different concepts in several sources. The first perspective for the term Autonomous Network Management (ANM) refers to a network management based on information technology that self-manages its own network operations with little or no need of human intervention, these operations include. detect, diagnose, repair failures, optimize performance and quality of service, Mortier & Kiciman (2006) Other sources define autonomic network management as a system where the network itself helps to detect, diagnose and repair failures as well as adapt its configuration and optimize its performance and availability, Mortier & Kiciman (2006). The same authors also pointed out that the autonomic network management techniques are not the first to use automated responses to configure and manage network. Previous automated network included IS-IS / OSPF, Mortier & Kiciman (2006)
Autonomic computing is also similar to autonomous network management and can be defined as the computer components that regulate and manage themselves in areas of configurations, healing, performance and protection to achieve the network goals. Traditionally network management activities have always been carried out by network managers or engineers. Today's network technology has grown very complex and mobile thus making the network management activities by the network managers very difficult, too costly, time consuming and error prone, plus these affect the availability of network application and services. Autonomous network management system can also resolve problems with minimal human intervention. According to IBM (2005)
3 The Goals for Autonomous Network Management
Autonomous network management goals are also categorized as according to IBM (2005) and these include;
The autonomous network management system has the ability to dynamically configure itself. Autonomous network management systems have the capability to dynamically and automatically add new nodes as they are replaced on the network and this dramatically leads to the scalability of the network, IBM (2005)
Automatically discover, diagnose and act to prevent any disruption as well as correction of faults. This action may lead to the system and subsystem to alter its own state or influence changes in the other elements of the environment. These changes are made to reduce or to help to eliminate the impact of failing components. There are a variety of faults that are manually hard to diagnose and this reduces the potential human errors, IBM (2005)
Self-Optimizing refers to the automatic monitoring and control of resources to ensure the optimal functioning with respect to the pre-defined requirements. Self-optimization helps provide a higher standard of service for both the system's end users and other external customers. The network deals with a massive amount of data that has been collected over a long period of time, the autonomous network system can use recognition techniques to analyze and anticipate where the additional capacity is needed. Self-optimizing networks can also anticipate, identify and protect against any threat anywhere in the network system, IBM (2005)
Autonomous network management aims to automatically control access to network resources and only allow right to users with privileges to access the right information and prevent unauthorised users or hackers from attacking the network. In order to achieve this goal autonomous network management system uses the analyzing technique. This can be obtained by analysing the data collected from repeated unauthorised login and suspicious communication patterns on network that might indicate ongoing attacks. In most cases the network system can shut down, lock or restart itself automatically. Self-protection consistently enforces security on the network and this reduces the administration security costs, IBM (2005)
4 Literature review focusing on the Autonomous network management techniques as applied to the network and how it differ from the traditionally network management.
In this chapter the research will critically analyse a literature that focus on autonomous network management techniques as applied to networks and where they differ from the normal management methods. The review will therefore, focus on the both autonomous and traditional techniques:
Autonomous network management
An Autonomous Network Management is defined as self-management system with the capability of self-configuring, self-healing, self-optimising and self-protecting of the network. Autonomous network management has the goal of increasing reliability, scalability and performance while reducing management and maintenance costs using various automated techniques. The autonomous network management system deploys AI (artificial intelligence) techniques that manage the adaptive behaviours of network devices and software. Lewis, O'Sullivan, Keeney, (2005)
Mobile agent refers to the software agents that have the capability of moving between locations and can make their own decisions according the change in the environment. These agents also combine their effort and solve any network threats that might cause the network to be inefficient Mortier, Richard, Kiciman, Emre (2006)
The autonomous network system employs mobile agents to self-manage devices and software that are added to the network node without the intervention of the network engineer. Example of the network devices are modems, routers, printers, antivirus, clients and servers etc. Bieszczad, Pagurek and White (1998)
It is this author's opinion that autonomous networks management is more efficient in performing this task because of the automatic nature with which it self-configures that allows the system to continue giving optimum service without affecting the end users.
Advantages of Autonomous network management (ANM)
Autonomous Network Management systems improve response times when request for change on the network are faster responded to which reduce the network downtime. Traditional managed networks, the request to the network change has to be collected, open the problem record before responding back. This process is most of the time proved to be unreliable.
Reduce skill requirements and affordability
Autonomous Network Management are configured in such a way that the network tasks are encoded within the system and tasks are automatically done rather than the network engineer performing that task in managed networks. This helps to reduce the level or degree of skills required to perform these network tasks. Traditional network management requires skilled network engineers to be on site all the time and this make traditional network management more expensive.
Automation of manual operations
Autonomous network management system dynamically responds to the request instantly based on information derived from the system. This helps to reduce the manual labour and the time required to respond to critical situations. Traditional network management uses ticket systems that are sent to the central server for the network engineer to track problems in the network and this tends to be time consuming.
Better management control
Autonomous network management system responds to requests rapidly which helps to solve network problems. Quality of service includes bandwidths, delay, error rates and availability. Traditional network management tends to take time responding to request since all the incidents have to be collected and then open problem record before rectifying the problem. This reduces quality of services and network availability if the network has to stop working.
Reduce maintenance cost
Autonomous network management reduce costs on maintenance since all the error and faults are dealt with in the system automatically by the system with a limited or no human intervention. Traditional network management needs skilled network engineers to sort out the network faults and errors manually. Hiring skilled IT profession can be costly.
An autonomous network management system tends to solve network faults by using the proactive solution. With the proactive solution network faults are detected before the network is degraded and this solution keep the network running and available all the time. Traditional network management deploy reactivity solution most of the time to overcome any fault in the network which mean that the network has to shutdown before the fault is detected and solved hence making the network to be unstable and unavailable all times. Reactive solution relays on manual prompting compared to proactive which is automatically prompted.
Traditional network management
Traditional network management systems are managed by a single management system server where data is stored, supported and executed. Since the entire network functions are located on the single server the network engineer can lock out the system and only the authorised user are the only allowed to access the network data. The entire network rely on the single machine and the failure of the management machine can lead to the collapsing of the entire network or part of the network being left unmanaged which can result into security threat and weakness. Physical data backup should be maintained all the time. The network engineers can remotely communicate with simple network management protocol (SNMP) agents which reside within the managed network nodes. All the applications are shared by every user on the network. The SNMP agents luck the capability of make their own decisions instead send alerts to the engineer who in turn do the tasks manually. This makes the performance, scalability, reliability of the network inefficient. (T. Chen, S. Liu)
5 Critical comparison Autonomous network management against the normal
There are various comparisons between autonomous network management and the normal network management. These comparisons are based on the strengths and weaknesses in terms of scalability, reliability, optimisation and security management
The autonomous network management system use mobile agents. Mobile agents that have the capability of moving between locations and can make their own decisions according the change in the environment. These agents automatically configure, protect, heal and optimise the without the intervention of the network engineer and this makes the autonomous network management very dependable on the larger networks. Traditional network management system use SNMP architecture model, the SNMP model is commonly used by network engineers to remotely manage network devices. SNMP uses agents that reside within or on the managed network devices to collect data that is passed on SNMP protocol to the network engineer. The SNMP agents are static and have their base located on or with the managed device and wait for the decisions made by the engineer who in turn physically manage the network tasks manually. Traditional network management is best for small networks.
Autonomous network management employs the (AI) artificial intelligence technique to proactively diagnose any fault that can lead to the network degradation without human intervention on the system. These faults could be software or network hardware that can cause failure of the network. Autonomous network system has the capability of deploying the correct procedure to element the cause of the fault. These procedures are automatically composed and planned by the AI agents.
Traditional network management use agents that reside within or on the managed devices to send alert or set alarms off in order to inform the network engineer about the fault of software or hardware on the network. Most of the times the network engineer collect information about the network but this sometimes can be very challenging if it become too complex networks. The network engineer identifies and diagnoses the fault manually. This method is sometimes very slow and time consuming that can result in the down time of the network for long period of time.
Autonomous network management self monitor the performance of the network and dynamically tune itself in response to the workload to improve overall utilization. That's to say if there is a network over load the autonomous network management system can automatically relocate resources.
Traditional network management deploys performance management techniques, monitoring tool detect and examine the network trend before being overloaded. The network management system will then send a massage or an alert to the network engineer instructing him to find solution before the network threshold exceeds the value set, Leinwand and Fang (2000).
Autonomous network management dynamically protects the network by using self-protection technique. The system prevents unauthorized access to the sensitive file and network resources. The security system monitors the users who have no authorization to access the network resources denied access. This can be obtained by analysing the data collected from repeated unauthorised login and change of the environment. Mortier & Kiciman (2006) the same author also pointed out that automated system can only handle common fault they are designed or programmed to deal with.
Traditionally the security monitoring tools are used to limit user from accessing sensitive files and network resources. An attempt to access unauthorized files or resource the engineer is notified and this done by the security management software tools. The engineer has the ability to lock the system away and only manually give access to the user with the rights.
This work has demonstrated through analysis of Autonomous Network Management (ANM), comparisons of ANM against traditional network management systems and a brief review of literature related to ANM and network management as whole that ANM as network management approach has several benefits that make it a more viable option to traditional/ engineer managed networks. While the advantages have been clearly outlined this work has also succeed in bringing forward potential draw backs that may arise from the universal adaptation of ANM in the management of all IT networks. In conclusion this author would advocate for the establishment of a framework or management system that adopts all the positives of ANM and combine them with those qualities of traditional/ engineer managed networks that provide solutions to the shortcomings of ANM.
- Jawahar & Elango (2001)"IFIP International Symposium on Integrated Network Management"
- MahmoudQ. H (2007) Cognitive networks: towards self-aware networks, John Wiley and sons ltd
- http://www.cis.udel.edu/~mills/autonet.html visited on 13/10/2009 at 0433
- http://en.wikipedia.org/wiki/Self-management_%28computer_science%29 visited 14/10/2009 at 2345
- http://en.wikipedia.org/wiki/Autonomic_Computing visited 14/10/2009 at 2345
- http://www.wipo.int/pctdb/en/wo.jsp?wo=2004046953 visited 18/10/2009 at 1919
- http://www.infc.ulst.ac.uk/informatics/cie/DANMS-Ericsson-Workshop.html visited 20/10/2009
- https://h10078.www1.hp.com/cda/hpms/display/main/hpms_content.jsp?zn=bto&cp=1-11-15-119_4000_0&jumpid=TC|14779|network%20management||S|b|4479726935 visited 20/10/2009
- IBM "an architectural blueprint for autonomic computing." ( 2005), White
- IBM, "An Architectural Blueprint for Autonomic Computing", April (2003)
- S. Bouchenak, F. Boyer, D. Hagimont, S. Krakowiak, A.Mos, N. Palma, V Quema, J. B. "Stefani Architecture-Based Autonomous Repair Management"
- M. Cheikhrouhou and J Labetoulle When Network Management Agents Become Autonomous, page 5-6
- T. Kornel (1992) Communication Network Management Prentice-hall International (UK) limited London 2nd edition
- N. Agoulmine, S. Balasubramaniam, D. Botvitch,J. Strassner, E. Lehtihet and W Donnelly "Challenges for Autonomic Network Management"
- Mortier, Richard , Kiciman, Emre(2006) " Applications, Technologies, Architectures, and Protocols for Computer Communication"
- K.Balos, M.Jarzab, D.Wieczorek and K.Zielinski (2007)"open interface for autonomic management of virtualized resources in complex system-construction methodology"
- A. Bieszczad, B. Pagurek and T. White(1998) "Mobile Agents for Network Management"
- D .Gürer, I. Khan, R. Ogier "An Artificial Intelligence Approach to Network Fault Management"
- A Leinwand and K Fang (2000)"network management A practical perspective" addition Wesley Longman page 95-180
- K. Ripon and K. Talukder (2004)"Intelligent Systems for Network Management An Alternative to the Centralized Approach of Computer Network Management"
- P. Astithas, G. Koutepas, B. Maglaris (2002) "Integrating Intrusion Detection and Network Management"
- J. Keeney, K. Carey, D. Lewis, D. O'Sullivan, V. Wade ( 2005)" Ontology-based Semantics for Composable Autonomic Elements"
- Mortier & Kiciman (2006)"Autonomic Network Management some Pragmatic Considerations"