Why ICT Fails and What CeDICT Will Do About It Boudewijn R. Haverkort University of Twente NIRICT Strategiedag, Utrecht February 12, 2009 Centre for Dependable.

Why ICT Fails and

What CeDICT Will Do About It

Boudewijn R. HaverkortUniversity of Twente

NIRICT Strategiedag, UtrechtFebruary 12, 2009

Centre for Dependable ICT Systems

What is a dependable ICT system?An ICT system is called dependable when

reliance can justifiable be placed on the services it delivers

A service delivered by an ICT system (in its role as provider) is its behavior as perceived by its users

But ICT systems do fail!

Why do ICT systems fail?

Hardware/physically-induced:• decreasing feature size in hardware components• shift from wired to wireless communicationSoftware/design-induced:• ICT systems are becoming distributed systems, in

which the interaction often causes difficulties• ICT systems are more and more software-driven,

and software is notoriously not error-freeHuman & operator induced:• ICT systems are not only used by experts anymore• some humans strive for deliberate failures (attacks)Natural & environmental disasters

Why are ICT failures a problem?They form a threat to humans and resources:• mission/operation-critical ICT, e.g., in space

or aircraft, but also in modern cars (fly/drive/brake-by-wire), and medical systems, e.g., software-driven radiology equipment

• commercially-critical ICT, e.g., e-bay, amazon, with high estimated cost of downtime:- e-bay: 225 K$/hr;- credit-card authorization company: 2.6 M$/hr

• infrastructurally-critical ICT forms backbone of most of our critical infrastructures:

- electrical power grid & water supply - public and freight transport systems

What will CeDICT do about this? The mission of CeDICT is to develop the methods, techniques and tools necessary to design and implement truly dependable ICT systems under cost and resource constraints

The “fab four” dependability “means”• fault avoidance: prevent the introduction of

faults by improving the design process• fault tolerance: adding redundancy (time,

space, information) to avoid service failures• fault removal: find faults (testing, probing,

model checking) and remove them, in design and implementation, as well as under operation

• fault forecasting: predict faults and future occurrences, through modeling, maintenance and rejuvenation

• choice depends on many factors!

CeDICT activities so far

• has become a partner in the JRU with ESI

• core of CeDICT took part in the Smartmix proposal on Dependable Embedded Systems with ESI

• workshops in Eindhoven and Utrecht (twice)

• partly active in FES2009 ICT proposal

• locally, new positions are being filled in

• CeDICT is not a brand-name yet!

New CeDICT activities needed!

• filling all six new chairs at the 3 TU’s

• become active in the JRU with ESI

• activity needed for 24x7 e-business dependability(partners needed from network and service providers and application fields)

• activity needed for critical ICT-driven infrastructures(partners needed from infra-providers, infra-users and ministeries, and ESI (via JRU))

• activities needed for dependable embedded systems with ESI (and in the JRU)

• partly covered in FES2009 ICT proposal

Conclusions

• Dependability of ICT systems is a key economical and societal concern

• ICT is everywhere, so chances abound!• CeDICT brings together the expertise & focus

to address the design of dependable ICT systems from all necessary angles

• But, it is the CeDICT people itself to fill the CeDICT skeleton with meat

• Original CeDICT-means have been allocated now, although in different ways at the three TU’s

• It is reasonable that OC&W expects to have concrete results out of CeDICT!

Actions needed to make CeDICT real!• Setting up joint research projects, really under

CeDICT umbrella• Work in cooperation with 3TU partners and ESI

for the JRU: research agenda, large national programs (e.g., STW Perspectief), EU-activities, etc.

• Publish in key dependability venues like IEEE DSN (www.dsn.org) and IEEE TDSC

• Publish, also with CeDICT affiliation, in topical conferences/journals (e.g., on verification, critical infra, embedded systems, etc.)

• Increase CeDICT awareness between/at 3TU sites• Reconsideration of the groups to be associated

with CeDICT

3TU.CeDICT people

• 3TU.NIRICT/CeDICT director: ApersNIRICT-MT: Haverkort, Corporaal, Lagendijk

• University of Twente: Haverkort (Depend. Comm.), Van de Pol (Formal Meth.), Hartel (Security), Nauta (IC design), Scanlon/Bentum (SRR)

• TU Eindhoven: Van der Aalst (Info. Syst.), Groote (Design. Syst.), Corporaal (Emb. Syst.), Koonen (Broadband Comm.), Etalle (Security), Liotta (Networking)

• TU Delft: Van Deursen (Softw. Eng.), Lagendijk (Info. Th.), NN (Comp. Arch.), Van der Veen (Circ. & Syst.), NN (Dep. Ad hoc Netw.), NN (Dep. Multimedia)

CeDICT kick-off

April 24, 2009, Utrecht

Coordination:Jaco vd Pol, Arie van

Deursen, Harold Weffers

• 9.30-10.00 Ontvangst• 10.00-10.15 Opening: Peter Apers (NIRICT)• 10.15-11.00 Keynote: Bill Sanders (UIUC)• 11.00-11.15 Break• 11.15-11.40 CeDICT: Marielle Stoelinga (UT)• 11.40-12.05 CeDICT: Twan Basten (TUe)• 12.05-12.30 Poster pitches• 12.30-13.40 Lunch @ posters• 13.40-14.40 Three NIRICT Lab Presentations/Demos • 14.40-15.05 CeDICT: William Scanlon (UT)• 15.05-15.30 CeDICT: Andy Zaidman (TUD)• 15.30-16.00 Break• 16.00-16.10 JRU ESI-CEDICT: Boudewijn Haverkort

(ESI)• 16.10-16.35 ESI: tbd• 16.35-17.20 Industrial Keynote: tbd• 17.20-18.00 Closing & drinks!