Emanics WP9 - Autonomic Management

Phase 1

Dynamics and automation will form the core of tomorrows management solutions. Their support needs new models, algorithms and frameworks. WP9 will address issues of autonomic management in the context of both fixed Next Generation Networks (e.g. QoS-enabled) and ad-hoc/ubiquitous environments. It will develop algorithms, frameworks and technology solutions for the autonomic/automated management of both types of networks.

WP9 will address issues of autonomic management in the context of both fixed Next Generation Networks (e.g. QoS-enabled) and ad-hoc/ubiquitous environments. It will develop algorithms, frameworks and technology solutions for the autonomic/automated management of both types of networks.

Motivation

Autonomic Management has recently emerged as an evolution of automated management. In terms of management functionality, systems can be characterised as unmanaged, managed, predictive, adaptive, autonomic. Adaptive closed-loop automated management is the first form of autonomic behaviour, with the ultimate target being fully distributed adaptive automated management, with relevant functionality built into the managed elements themselves. In this light, WP9 tries to integrate relevant know-how and approaches for both fixed/cellular and ad hoc/ubiquitous environments, answering questions and identifying key further research issues.

Relevant work is split into two parts: frameworks and technologies. Integration work in the area of frameworks addresses adaptive automated management approaches for fixed/cellular networks and distributed adaptive management for ad hoc / ubiquitous environments. Integration work in the area of technologies addresses policy-based management on one hand and emerging XML-based technologies and in particular Web Services as the unifying technology for future autonomic management systems.

The participants have expertise and have worked / are working in other research projects in all these areas. They complement well each other in the sense that they address various sub-problems from different perspectives. A key target will be the enumeration, harmonisation and integration of existing approaches addressed by the various participants and the identification of key further issues. The complementary expertise of the WP9 participants is evident from the description of their intended contributions that follow.

Objectives

This work package is part of the Research Activity and will address issues of autonomic management in the context of both fixed Next Generation Networks (e.g. QoS-enabled) and ad-hoc/ubiquitous environments. The key objective of this work package is to develop frameworks for the autonomic/automated management of both types of networks focusing on the required functionality that need to be present in such frameworks as well as the management technologies used to support the requirements of autonomic management of fixed and ad-hoc networks. In this context, the following aspects will be addressed:

• Automated Management of fixed Networks:
  • Develop an overall automated management framework that will target resource optimisation and service fulfilment based on Traffic Engineering (TE)
  • Identify offline and dynamic traffic engineering approaches that can produce an optimal configuration of the network based on predicted traffic demand and network monitoring information
  • Consider scalable monitoring approaches to support configuration evolution and scalable collection of relevant network information
• Autonomic Management of ad-hoc/ubiquitous environments:
  • Develop an overall framework identifying the components that need to be present in order to support self-management of ad-hoc networks
  • Investigate dynamic clustering algorithms so that cluster heads act as local managers and policy decision points as well as programmable approaches to support node capabilities alignment – self-organisation
  • Investigate methodologies to support identification of malicious nodes that attempt denial of service attacks – self-protection
  • Context gathering, representation and dissemination, taking also into account resilience and efficiency, so that relevant decisions on self-protection and self-organisation are adaptive to it – self-adaptation
• Consider Policy-based Management as the management paradigm used to manage both fixed and adhoc networks and investigate issues such as policy refinement and conflict detection and resolution
• Investigate the use of emerging technologies for management of fixed/ad-hoc networks based XML-based technologies and in particular Web Services as the key unifying technology.

Tasks

In order to achieve these objectives, the following tasks have been defined:

T9.1: Management frameworks

The work in this work package will start with the definition of the overall management frameworks for both fixed QoS-enabled IP and ad-hoc networks. The relevant components and their interactions will be specified supporting the required functionality to achieve autonomic management.

T9.2: Algorithms for autonomic management

In the context of management of fixed QoS-enabled IP Networks, traffic engineering algorithms will be studied that aim at enhancing the network performance while at the same time optimizing the use of network resources. Issues such as resilience and fault tolerance will also be incorporated in the proposed algorithms. The TE approaches will cover both offline (time-dependent) and dynamic (network-state dependent) network optimisation for IP and MPLS-based networks. A sophisticated network monitoring mechanism will be developed so that the configuration of the network can be self-adapted with fast convergence time according to the data obtained from the monitoring system. The monitoring system should be able to report promptly, so that online TE mechanisms can be triggered for traffic re-optimisation.

In the context of autonomic management of ad-hoc networks, algorithms and mechanisms that will support self-* functionality. In order to support self-capability-alignment, we will study programmable mechanisms for node capability alignment, with required protocols and services diffused and activated; the decision on what to deploy will be based on application/service requirements, context and policies. Self-organisation will be addressed by specifying algorithms for optimally splitting the network into local regions to be managed (“clusters”) with cluster heads cooperating for overall cooperative network management. Security (selfprotection) issues will also by addressed by specifying methodologies for identifying malicious nodes that attempt to flood the network by observing traffic patterns, or simply identifying nodes that act in a selfish fashion by not cooperating e.g. not routing other nodes traffic. Finally, we will study algorithms for efficient context information gathering, processing/aggregation and dissemination, taking also into account resilience and efficiency, so that relevant decisions on self-protection and self-organisation are adaptive to it (self-adaptation).

T9.3: Alternative paradigms & supporting technologies

Policy-based Management will be considered as an alternative management paradigm for both fixed and adhoc networks where network configuration will be expressed as a set of policy rules. In this case, it is becoming increasingly important to perform analysis of these policy specifications in order to detect and resolve potential inconsistencies. The policy/configuration analysis approaches will be based on reasoning techniques as identified in the literature. Finally this work package will examine XML-based technologies and in particular Web Services as the key unifying technology of the developed autonomic management frameworks. We will identify the key technological issues that have prevented integrated automated management until now and investigate relevant approaches and mechanisms, including selective and bulk data retrieval, data modelling to support scalability, scalable event management, transaction support for configuration management, including both object and XML-document based approaches and finally security, including authentication/authorisation and access control.

Partners

The following EMANICS partners are involved in this work package:

• INRIA LORIA (INRIA) (Olivier Festor)
• University of Twente (UT) (Aiko Pras)
• International University Bremen (IUB) (Juergen Schoenwaelder)
• Imperial College (IC) (Emil Lupu)
• Universidad Polytechnica de Catalunya (UPC) (Joan Serrat)
• University of Federal Armed Forces Munich (CETIM) (Gabi Dreo-Rodosek)
• Ludwig-Maximilian University Munich (LMU) (Helmut Reiser)
• KTH, Royal Institue of Technology (KTH) (Rolf Stadler)
• Oslo University College (HIO) (Mark Burgess)
• University of Surrey (UniS) (George Pavlou)

Deliverables

The deliverables produced by this workpackage can be found here.

Phase 2

Objectives

This work-package is part of the research activity and will address issues of autonomic management. Building on the collaborative work performed by the participating partners in the first phase, the second phase will identify emerging approaches for autonomic management,study them and propose solutions to tackle potential problems. In parallel, one of the key objectives of this work-package will be to work on certain aspects of autonomic management that were introduced in the first phase. In this respect, the following aspects will be addressed:

• Intra and Inter-Domain Autonomic Management of Fixed Networks
• Automated inter-domain QoS management
• Self-management of MPλS optical networks
• Automated intra- and inter-domain fault-management i.e., service recovery, resilience
• Autonomic management of ubiquitous environments
• Context-awareness for self-management of ubiquitous environments using ontologies
• Adaptive distributed management (i.e., incorporating management patterns in cfengine)
• Distributed self-management for wireless and mesh networks
• Policy-based management
• Monitoring, policy representation, refinement, analysis etc.
• Emerging paradigms for autonomic network management
• Fundamentals of P2P for network management (i.e., incentives, collaboration)
• P2P technologies for network management
• Management of P2P environments

WP9 will follow an open-call model to select joint proposals and allocate funds accordingly.

Description of work

T9.1: Intra and Inter-Domain Autonomic Management of Fixed Networks

Building on the frameworks for autonomic management of fixed networks presented in the first phase of this work-package, work for the second phase will expand the research realm to address the complexities involved in Inter-domain issues. A case study of inter-domain QoS automated management will receive collaborative interest. Furthermore, intra-domain issues will continue to be investigated. A case-study of particular interest, namely optical networks and their self-management, will be studied. Finally, automated intra- and inter-domain fault management for service recovery and resilience is another research aspect of this task

T9.2: Autonomic management of ubiquitous environments

Anomaly behaviour in monitoring context information will be one of the topics of interest for this research task. Context management will also be addressed, with the focus being on ontologies describing the context information. The work that begun in the first phase on distributed autonomics by incorporating management patterns in cfengine will be expanded to cover the wider area of adaptive distributed management. Correspondingly, distributed self-management of MANETs that was presented in the first phase will continue to be addressed, incorporating studies on wireless and mesh networks to better cover this research domain.

T9.3: Policy-based autonomic management

Policy-based management will be considered as an alternative management paradigm for both fixed networks and wireless ubiquitous environments, considering the specific requirements of each environment. Network monitoring to provide accurate policy triggering information is important, while policy representation, refinement and analysis are of paramount importance in order to detect and resolve potential inconsistencies in dynamic co-operative environments.

T9.4: Peer-to-peer approaches for autonomic network management

There has been an increasing research interest towards P2P principles and technologies for autonomic network management. The partners of EMANICS are collaboratively going to be working on various aspects of P2P autonomic network management.

Partners

The following EMANICS partners are involved in this work package:

• INRIA LORIA (INRIA) (Olivier Festor)
• University of Twente (UT) (Aiko Pras)
• International University Bremen (IUB) (Juergen Schoenwaelder)
• Imperial College (IC) (Emil Lupu)
• Universidad Polytechnica de Catalunya (UPC) (Joan Serrat)
• University of Federal Armed Forces Munich (CETIM)(GabiDreo-Rodosek)
• Ludwig-Maximilian University Munich (LMU) (Helmut Reiser)
• KTH, Royal Institue of Technology (KTH) (Rolf Stadler)
• Oslo University College (HIO) (Mark Burgess)
• University of Surrey (UCL) (George Pavlou)
• University of Zurich (UNIZH) (Burkhard Stiller)

Deliverables

The deliverables produced by this workpackage can be found here.

Phase3

Objectives

This work-package is part of the research activity and will address issues of autonomic management. Building on the collaborative work performed by participating partners in the previous two phases of the project, the third and last phase will drive the work to completion. Aiming to propose a comprehensive set of solutions to the many challenges surrounding autonomic management, the work will extend previous efforts to finalise open issues and tackle newly identified research directions. In this respect, the following aspects will be addressed:

• Autonomic management of fixed networks and ubiquitous environments
  • Orchestration of autonomic management elements across domains
  • Collaborative models for automated inter-domain service/resource negotiation
  • Automated intra- and inter-domain fault-management i.e., service recovery, resilience
  • Security management of wireless sensor networks
  • Energy efficient MANETs

• Policy-based autonomic management
  • Vulnerability management of autonomic systems
  • Self-configuration of limited capability devices
  • Policy continuum and mechanisms for autonomic management

• Peer-to-peer approaches for autonomic network management
  • Automated handling of malicious behaviour
  • Joint optimisation of overlays and network infrastructure
  • Alignment of local goals with system-wide objectives

WP9 will follow an open-call model to select joint proposals and allocate funds accordingly.

Description of work

T9.1: Autonomic management of fixed networks and ubiquitous environments

Building on the foundations established in previous phases, this task will extend the efforts to provide solutions for inter-domain management in fixed networks and address issues in resource constrained ubiquitous environments. Federated management between administrative domains will encompass the orchestration of several autonomic management elements for intelligent service composition and end-to-end service delivery. Part of this work will involve the development of automated service and resource negotiation models between domains, which can allow for the establishment of mutually beneficial agreements among collaborating agents. Another aspect that will continue to receive attention is the automation of service recovery and resilience processes in the context of fault management. While previous work in ubiquitous environments mainly focused on effective context management, the third phase will address efficiency and optimisation issues in resource usage of small networked devices with limited capabilities. Examples include self-healing and protection mechanisms to prolong the operational readiness of wireless sensor networks, and the development of self-configuration mechanisms to optimise processing and communication overheads on handheld devices forming MANETs.

T9.2: Policy-based autonomic management

As in the previous phases of the project, policy-based management will continue to represent an alternative management technology enabling autonomic capability in both fixed networks and wireless ubiquitous environments. More specifically, the policy paradigm will play a key role in a number of WP9 activities and will be used to investigate how autonomic systems can manage their own vulnerabilities from a security perspective, to guide the self-configuration of networked devices with limited capabilities, and to automate the decision making process in the context of fault recovery planning. Furthermore, the specification of a policy continuum and the implementation of relevant mechanisms will be part of an autonomic management system architecture.

T9.3: Peer-to-peer approaches for autonomic network management

Work on peer-to-peer principles and technologies for autonomic network management will be investigating misbehaviour and optimisation issues in collaborative environments. As the business models of P2P overlays mature, it is clear that there is ample opportunity for joint optimisation of the overlays and the underlying network infrastructure. This task will explore such an interaction using game theoretical models, in order to explore the possibilities for joint optimisation and incentive alignment that might arise. Furthermore, the alignment of local agent goals with system-wide engineering objectives will also be investigated including truthful revelation issues in the context of distributed P2P protocols. Lastly, this task will consider scenarios where malicious nodes are present in the network. Mechanisms and techniques will be developed to handle malicious behaviour in an automated manner so that the performance of applications does not suffer.

Projects active during 2009

Automated Service Management using Emerging Technologies (ASMET3)

One of the key challenges in today’s IT environments is automatically determining and applying recovery solutions in response to emerging faults. ASMET3 extends previous efforts in this area by considering peer-to-peer technologies and artificial intelligence techniques. The knowledge discovery and fault resolution activities focus on a distributed case-based reasoning system that exploits various on-line knowledge sources and reasoning capabilities in a decentralized, self-organizing platform provided by peer-to-peer technologies. It also looks into reasoning methods that can deal with semi-structure fault data in order to provide useful information for fault resolution. The focus is on a reasoning method that can take advantage of the textual description of fault data such as bug reports extracted from bug tracking systems.

Federated Management in the Future Internet (FeMaFI2)

In an environment where different autonomous entities will be sharing the same set of resources, federation mechanisms need to be in place such the resources are used effectively towards specific objectives. FeMaFI2 builds upon the architecture developed in the previous phase of the project that supports governance and negotiation functionalities. The objectives of this activity for the year 2009 were established as follows:

* Specification of policies for the orchestration of Autonomic Management Systems (AMSs).

* Design of algorithms and methodologies for the orchestration of AMSs based on the defined policies.

* Definition of mechanisms and protocols to establish negotiation between multiple service providers in competitive environments.

* Enhancement of the negotiation approach with automated resource negotiation models based on game theoretic approaches.

Policies for Autonomic Management (PUMA)

The network’s design is moving towards a different level of automation, autonomicity and self-management. A key element in the above conception is the elementary autonomic cell; this is, the minimum functionality to deploy exhibiting self-management properties. Our vision is that sets of these elementary cells would then be coordinated (orchestrated) to make a complete system or network self managed. This elementary cell is called Autonomic Management System (AMS) in the context of this work. We assume that the behaviour of an AMS is policy-based and the purpose is to investigate and provide a solution for the set of policies to drive the behaviour of AMSs.

Autonomic Management of Limited Capability Devices (AMICAD)

Wireless Sensor Networks (WSNs) are closely related to mobile ad-hoc networks due to sharing the same general problems. Nonetheless, there are some differences between the two concepts, mainly with respect to energy considerations. In contrast to mobile ad-hoc networks, WSN nodes have fewer resources such as non-rechargeable batteries, less memory, slower CPUs and so on. The lifetime of a WSN can be shortened rapidly by misbehaving nodes or attacks on the network. Security aspects such as self-protection help to increase the operational readiness and secure the network against attacks. Former discussed concepts focused on self-configuration due to the complexity of large scale networks. To follow up on these, AMICAD investigates self-management concepts for WSNs in the context of security.

Distributed and Autonomic Real-Time Traffic Analysis (DATTA3)

During the last phase of the DATTA3 activity the two main areas which were investigated were: 1) how to distribute traffic monitoring data for distributed processing of such data, and 2) how well can traditional centralized relational databases cope with high amounts of traffic data. In addition, a separate investigation of how SNMP traces are actually used in real networks was performed.

P2P Collaboration and Storage (P2P COST2)

This project proposes a protocol for Sybil-aware directed swarming for the distribution of live content. This mechanism organizes peers to reduce delay and increase the number of stream recipients without a central organizer, even in the presence of churn and unreliable users. The system is resistant to certain classes of known mechanisms for "lying peers" and Sybil attacks. Peers are rewarded for their contributions to uploading the stream by moving closer to the peercaster, giving them both increased liveness and stability. Although Sybil attacks are still possible, they produce a very limited benefit for the attacker.

Partners

The following EMANICS partners are involved in this work package:

• University College London (UCL)(George Pavlou)
• INRIA LORIA (INRIA) (Olivier Festor)
• University of Twente (UT) (Aiko Pras)
• Jacobs University Bremen (JUB) (Juergen Schoenwaelder)
• Oslo University College (HIO) (Mark Burgess)
• University Politecnica de Catalunya (UPC) (Joan Serrat)
• University of the Federal Armed Forces Munich (CETIM) (Gabi Dreo-Rodosek)
• Ludwig-Maximilians University Munich (LMU) (Helmut Reiser)
• University of Zurich (UNIZH) (Burkhard Stiller)
• Royal Institute of Technology (KTH) (Rolf Stadler)
• Imperial College (IC) (Emil Lupu)

Deliverables

The deliverables produced by this workpackage can be found here.