1 / 11

Dependable, Self-Adaptive, Self-Healing, Distributed Systems through Reflection

Dependable, Self-Adaptive, Self-Healing, Distributed Systems through Reflection. Gordon Blair and Geoff Coulson, Lancaster University, UK Research track to be carried out in collaboration with Jean-Charles Fabre, LAAS, France Jean-Bernard Stefani, INRIA, France.

elewa
Télécharger la présentation

Dependable, Self-Adaptive, Self-Healing, Distributed Systems through Reflection

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Dependable, Self-Adaptive, Self-Healing, Distributed Systems through Reflection Gordon Blair and Geoff Coulson, Lancaster University, UK Research track to be carried out in collaboration with Jean-Charles Fabre, LAAS, France Jean-Bernard Stefani, INRIA, France

  2. Dependability and reflection • baseline • work on reflective distributed systems platforms at INRIA, LAAS, Lancaster (and elsewhere) • open up ‘black-box’ systems to inspect, adapt, extend • facilitates handling of non-functional properties • separation of concerns (‘base-level’ and ‘meta-level’) • reduces complexity

  3. Research goals • enhance the dependability of distributed systems by introducing self-adaptive, self-healing properties • examples • run-time inspection • e.g. instrumentation, debugging • run-time adaptation • e.g. hot module-update, dynamic reconfiguration/ fault repair • run-time extension • e.g. fault injection for testing, add fault-tolerance protocols • implement and evaluate in component-based architectures (OpenCOM, THINK, ...)

  4. Research issues • determining suitable component-based architectures for reflective self-adaptation/ self-healing • e.g. sensors, actuators and policies in meta-space... • sidestepping the dependability/ complexity trade-off • how to ensure the dependability of reflective meta-models themselves • what primitives need to be built into the component model itself (security? checkpointing? ...) • evaluating these architectures in dependability scenarios

  5. Industry perspective • focus on integrating existing techniques into innovative architectures (e.g. component-based structure, flexible interaction styles, ...) • not constrained by current standards • focus on recoverability and self-healing rather than on avoiding failure • explore innovative, self-aware, autonomic, architectures which are nevertheless ‘simple’ • component-based structure with self-awareness and self-adaptation • reduce complexity by separating concerns • work on wrapping existing systems • a NoE/IP successor to CaberNet should sponsor research in innovative component-based platforms as an approach to building dependable systems • multiple platform-related projects could be integrated by selecting a common component model as a common basis

  6. A Recommendation from CaberNet’s Links-to-Industry Forum Geoff Coulson and David Hutchison (Joint Chairs of the Links-to-Industry Forum) Distributed Multimedia Research Group, Department of Computing, Lancaster University geoff@comp.lancs.ac.uk

  7. Context of CaberNet’s Links-to-Industry Forum • ‘steering committee’ of senior industrial R&D people in Europe and US • role • commenting on relevance of CaberNet research to industry • making strategic recommendations for the future • recent meeting in London, 12th Nov

  8. Composition of the Forum • members present at recent meeting • Adrian Colyer, IBM Hursley, UK • Andrew Herbert, Microsoft Research, Cambridge, UK • Helmut Leopold, Telekom Austria • Dave Marples, Global Inventures, USA • Ian Marshall, BT Exact Technologies, UK • Andreas Mauthe, KIMK, Germany • Paul McKee, BT Exact Technologies, UK • Alberto Ruiz de Olano, Ikerlan, Spain • Chris Sluman, OpenIT, UK • Joe Sventek, Agilent, UK • other members (not at recent meeting) • Spyros Denazis, Hitachi, UK • John Evans, Marconi, UK • Michel Gien, Jaluna, France • Gerard Hartnett, Intel, Ireland • Stephen Hope, Orange, UK • Jens Kristensen, Ericsson, Denmark • Kathleen Milstead, France Telecom, France • Philippe Robin, ARM, UK • Dan Waddington, Lucent, USA • John Zinky, BBN, USA

  9. Observations • improving dependability is a crucial theme for future European research • in the area of large-scale distributed systems, high complexity is the biggest single factor hindering dependable systems development => high degree of human error • no need for research push in specific dependability technologies • fault tolerance, clustering, replication, etc.

  10. Recommendations (1 of 2) • focus on integrating existing techniques into innovative architectures (e.g. component-based structure, flexible interaction styles, ...) • not constrained by current standards • focus on recoverability and self-healing rather than on avoiding failure • explore innovative, self-aware, autonomic, architectures which are nevertheless ‘simple’ • component-based structure with self-awareness and self-adaptation • reduce complexity by separating concerns • work on wrapping existing systems

  11. Recommendations (2 of 2) • a NoE/IP successor to CaberNet should sponsor research in innovative component-based platforms as an approach to building dependable systems • multiple platform-related projects could be integrated by selecting a single component model as a common basis

More Related