Towards a Self-Organizing Model for Virtual Network Provisioning Masters Thesis Proposal Carolina Valadares and Carlos Lucena 2013/I
The Problem Network Ossification High dependence on human intervention for configuration and troubleshooting.
The Problem Network Ossification High dependence on human intervention for configuration and troubleshooting. Virtual Networks
Proposed Solution
Proposed Solution Physical Network
Proposed Solution Physical Network Physical Router
Proposed Solution Physical Network Physical Link
Proposed Solution Virtual Network
Proposed Solution Virtual Network Virtual Router
Proposed Solution Virtual Network Virtual Link
Proposed Solution Two main characteristics: - Adaptation - Physical Resource Sharing
Environment Changes Virtual Router Overload/ Virtual Router Failure
Environment Changes Unbalanced Virtual Links
Environment Changes Physical Router Overload/ Physical Router Failure
Proposed Solution Autonomic Agents
Proposed Solution Agent Communication
Self-Organizing Model Adaptive Plans: Replace Virtual Machine Live Migrate Virtual Machine Balance virtual link With and without the creation of new virtual machine Custom Control Loop (IBM extension): Collector; Analyzer; Decision-Maker; Norm Checker; and Executor.
Self-Organizing Model Self-Organizing Monitoring: Event-based and on demand; Dynamic adjustment of a set of parameters (Norms). Analyzing: State-based and history-based; Use of metrics; Uses up-to-date knowledge about its current status. Decision Making: Triggered in response to external or internal event; Apply the most appropriate decisions without any human support ; Adaptation rate. Norms Self-Tuning Reputation
Self-Organizing Model Self-Awareness Knowledge representation Structure knowledge Behavior knowledge Adaptive Plans Knowledge knowledge acquiring: (Inferred knowledge) Infers current virtual and physical network topology; Infers event execution; Infers network status; Implicit coordination. Discovering knowledge existence. Knowledge sharing Exchange messages only in the neighborhood.
Self-Organizing Model Norms/Reputation Self-Tuning: Dynamic adjustment of a set of parameters (minor adaptation operations – Control Loop parameter tuning) Reputation: To support the live migration of virtual routers, the decision maker takes into account the link Stress together with the Entities Reputation – popularity, rather than only Network parameters. History-based to describe the requests rate of a virtual/physical router.
Next Directions Reputation Self-awareness Experiments E01: Self-Organizing E02: Self-Organizing and Self-Awareness E03: Self-Organizing, Self-Awareness and Self- Tuning E04: Final Experiment with Self-Organizing, Reputation, Self-Awareness and Self-Tuning Cross-Validation E i vs. Baseline
Chronogram
References [1] C. Prehofer and C. Bettstetter, Self-organization in communication networks: Principles and design paradigms, IEEE Communications, [2] Z. Movahedi et al., "A Survey of Autonomic Network Architectures and Evaluation Criteria, Communications Surveys & Tutorials, IEEE, [3] Ines Houidi, Wajdi Louati, Djamal Zeghlache, Panagiotis Papadimitriou, Laurent Mathy, "Adaptive virtual network provisioning, Proceedings of the second ACM SIGCOMM workshop on Virtualized infrastructure systems and architectures, 2010.
Questions?