System level testing of virtual switch (re-)configuration over IP

Verified Software: Theories, Tools, Experiments

2008

Abstract. In this paper, we consider the Grand Challenge under a very specific perspective: the enabling of application experts without programming knowledge to reliably model their business processes/applications in a fashion that allows for a subsequent automatic realization on a given platform. This goal, which aims at simplifying the tasks of the many at the cost of ambitious and laborious tasks for the few, adds a new dimension to the techniques and concepts aimed at by the Grand Challenge: the application-specific design of platforms tailored for the intended goal. We are convinced that the outlined perspective provides a realistic and economically important milestone for the Grand Challenge. MotivationSince the very beginning of computer science, the mismatch between design for machines and design for brains has been a constant pain, and a constant driver for innovation: it was always clear that descriptions that are good for machine processing are inadequate for an intuitive human understanding, and that descriptions which are structured for easy comprehension contain a lot of 'syntactic overhead', that typically slows down their automatic processing.Compilers were designed to overcome, or better to weaken, this mismatch: Rather than trying to construct machines that work as humans think, it seemedmore appropriate to translate comprehensible descriptions into machine-adequate representations. Besides classical compilers that translate high-level programming language into machine code there are also other means of automated code generation from more abstract descriptions: to this group belong parser generators, data flow analysis generators, compiler generators, and all the modern OOtools that translate, e.g., UML descriptions into code fragments. Most drastic are those versions of Model Driven Design (MDD) that aim at totally replacing the need of programming for most application development using model construction. Thus the original desire to lift low-level machine code (prescribing How to run the machine) to more abstract programs (describing What to achieve) has become reality at increasing levels of abstraction from the hardware, and leading to extreme approaches like requirements-based programming, which aims at automatically making the user's or customer's requirements executable.

show abstract

Section: An Amdd-realizationmentioning

confidence: 99%

Section: Ite: the Integrated Test Environmentmentioning

confidence: 99%

Section: Ite: the Integrated Test Environmentmentioning

confidence: 99%

Section: Ite: the Integrated Test Environmentmentioning

confidence: 99%

See 2 more Smart Citations

From the How to the What

Verified Software: Theories, Tools, Experiments

2008

show abstract

“…Our direct industrial experience stems from the area of Computer Telephony Integrated (CTI) systems. In the past we developed a piece of automated testing equipment (ITE) [14,18] that has been used for industrial system-level testing of over 200 COTS applications that interoperate with a family of midrange telecommunication switches. Characteristic here was the absence of any form of (formal or semiformal) operational model for the hardware and software systems constitute a CTI scenario, which are therefore seen and treated as black boxes.…”

mentioning

confidence: 99%

Knowledge-based relevance filtering for efficient system-level test-based model generation

Raffelt²,

Innovations Syst Softw Eng

2005

Test-based model generation by classical automata learning is very expensive. It requires an impractically large number of queries to the system, each of which must be implemented as a system-level test case. Key in the tractability of observation-based model generation are powerful optimizations exploiting different kinds of expert knowledge in order to drastically reduce the number of required queries, and thus the testing effort. In this paper, we present a thorough experimental analysis of the second-order effects between such optimizations in order to maximize their combined impact. MotivationValidating complex heterogeneous systems escapes established formal approaches, both for system testing and for design verification. Characteristic here in fact is the lack of (formal or semiformal) operational models for many of the hardware and software systems constitute such a scenario. Typical application domains are telecommunication systems and systems on a chip, but also EAI (enterprise application integration) scenarios, which are large software integration projects suffering from an almost complete lack of usable models for the components' behaviors (think, e.g., of an SAP installation with all the customization modules!). This situation is not only due to a lack of care in the production and maintenance of up-to-date models of all the system components. Rather, there is a policy of hiding intellectual prop- T. Margaria (B) erty, since the operational models of the commercial products would reveal too much about protected techniques and designs. Thus the lack of models and the need to "discover" them postproduction will accompany system-level design and testing for a long time to come.Particularly typical in practice are systems that, additionally, include different commercial components that come from various producers and are made available as products, without any model. Our direct industrial experience stems from the area of Computer Telephony Integrated (CTI) systems. In the past we developed a piece of automated testing equipment (ITE) [14,18] that has been used for industrial system-level testing of over 200 COTS applications that interoperate with a family of midrange telecommunication switches. Characteristic here was the absence of any form of (formal or semiformal) operational model for the hardware and software systems constitute a CTI scenario, which are therefore seen and treated as black boxes. In particular, there is no basis for test coverage considerations, focused test suite enhancement, or systematic maintenance support.Fortunately one can observe that the models one needs to "discover" in order to profitably use the components are far smaller than the full model of the implementation: they are much more abstract, since they provide information about only the interface behavior exposed to the environment. In this respect, complexity experts often talk of cognitive complexity (i.e., what a user needs to know in order to use a product), which is much smaller than the technical or intensiona...

show abstract

Lightweight coarse-grained coordination: a scalable system-level approach

2004

STTT