1. Initiatives for FP6 A few thoughts on flip chart about: Open Infrastructures Paulo E. Veríssimo LaSIGe-FCUL, Univ. Lisboa pjv@di.fc.ul.pthttp://www.navigators.di.fc.ul.pt

2. Methodology Current situation Problems Problems  Challenges/Opportunities Hopefully: –Industry will see problems –Research will see opportunities Research in the good direction?

3. Situation More and more systems relying on “open network environments” (aka Internet), COTS components, unskilled users, etc. Even if systems use specific sub-systems… e.g., fieldbus networks, mobile, private subsystems …common feature will be “openness and interdependence generated through the Internet” Even so-called “critical infrastructures” are exposed to this undesirable syndrome economic constraints, widespread use of COTS components, business and organisational constraints

4. Situation Increase in the risk associated with using technologies that not adequate to the role they are playing More exposure of highly-critical systems More not-so-critical systems exposed

5. Exposure Faults/Attacks/Errors/Intrusions

6. COTS INSIDE!

7. Interdependence Interference, Error propagation

8. Pitfalls of modularity and distribution… A distributed system is the one that prevents you from working because of the failure of a machine that you had never heard of. [ L. Lamport]

9. Pitfalls of modularity and distribution… Since: –Machines fail independently, for a start –They influence each other, –They communicate through unreliable networks, with unpredictable delays...gathering machines renders the situation worse: –The reliability (<1) of a system is the product of the individual component reliabilities, for independent component failures –R(10 @ 0.99)= 0.9910= 0.90; R(10 @ 0.90)= 0.9010= 0.35

10. change technologies, or make them work better Problems change technologies, or make them work better?

11. Problems change technologies, or make them work better? general concern with dependability, survivability, trustworthiness, and so forth architecture, paradigms, design assessment and assurance, amongst other factors

12. Main Problem/Opportunity areas Applications that are critical (e.g. SCADA, CCC), supported by semi-open geographically dispersed infrastructures Applications that are commercial (e.g. B2B, B2C), supported by very open extra-powerful site-based server compounds Applications of a mixed nature w.r.t. purpose, but whose distinguishing feature is being supported by ad-hoc collections of wireless and mobile entities, immersed in active environments of ubiquitous and inconspicuous devices.

13. Opportunities Reconcile uncertainty with predictability Manage apriori undefined or evolving failure modes and system configurations Manage exposure and interdependence Handle operation mistakes, unskilled users Adapt fault/attack ranges: from script kids to cyber terrorists; from low to high power attackers; from harsh to bening environments

14. A perspective on possible research directions Conceptual framework for Dependability Under Uncertainty (DUU) Reference Model and Architectural Framework for Acceptable Availability under Attack (A 3 ) Architectural frameworks for Dependability and Adaptation (DepAd)

15. Research directions: a perspective Conceptual framework for Dependability Under Uncertainty (DUU) –New conceptual frameworks are needed to meet the new challenges for dependable systems. –Namely, is it necessary an evolution of the current concepts: –(a) in order to encompass not only the reliance the system can justifiably give (aka trustworthiness), but also the way to build that reliance in the system users (aka trust), and their precise relationship, and –(b) in order to encompass uncertainty and adaptation, while remaining precise (e.g., trust to the extent of…)

16. Research directions: a perspective Reference Model and Architectural Framework for Acceptable Availability under Attack (A 3 ) –Distributed Denial-of-service Attacks are extremely hard to counter with current state-of-the art technology. –One strong reason for it is that the problem has been tackled from an engineering perspective under current network and distributed system models. –We believe there is the need for a reference model encompassing the notion of Acceptable Availability under Attack, and an architectural framework supporting the concept, and that this constitutes a broad area of research.

17. Research directions: a perspective Architectural frameworks for Dependability and Adaptation (DepAd) –When dealing with uncertainty of the environment, hard to provide guarantees, such as performance, timeliness, reliability or security. –E.g. in ambient intelligence, active environments, sentient objects (home, air, space/land traffic, robot or enhanced human teams). –Current state-of-the-art uses heuristic QoS management based on aprioristic assumptions. –Need architectural frameworks for adapting functional and non-functional properties while at least providing guarantees on how dependably it is adapting

18. Where to find us Navigators Group at LaSiGE Univ. Lisboa, Portugal http://www.navigators.di.fc.ul.pt related research lines in the site: Fault and Intrusion Tolerance in Open Distributed Systems Timeliness and Adaptation in Dependable Systems Feel free to email: Paulo Veríssimo --- pjv@di.fc.ul.pt