Services Science and Services Computing

New models of computation such as cloud computing, Big Data, and the Internet of Things have fundamentally upended common assumptions about the nature and purposes of computation. One thing that may be said about these and some other such paradigms is that they almost always require computation to be provided as a service to some entity seeking a larger end, rather than regarding the computation as being an end in itself. Services also come with their own set of challenges; e.g., services are typically more difficult to create and manage well than product creation, and services almost always require humans in the loop in critical functions.

Yet, classical computing curricula and research directions give few insights into how students may be taught to understand how computation may be fashioned to work as a component in a service existing in a larger business or social context.

This talk considers certain problems and issues that arise in services computing that are not covered well in standard discussions of computing technologies.

Safety Issues in Cyber-Physical Systems

Computers and software components are now integral to large complex systems in almost all sectors of modern society. The safety of such systems is therefore dependent upon many components with diverse technologies, and their interactions, all working properly. However, complex systems are fundamentally difficult to deal with: they often run in degraded mode, with minor component errors being commonplace and routinely accepted; they are also constantly changing, with no complete picture of the system state available at most times. Economic and social pressures, as well as concerns about other system performance aspects such as availability and dependability, etc., tend to influence system design and operation to a large extent.

This talk aims to give a broad introduction to the study and practice of safety in the design and operation of complex cyber-physical systems. The field is too vast to be covered in a single talk, but a representative overview of the many issues concerning safety is given to highlight some of the important aspects that may be considered to be worth further study. The topic may be of interest to practitioners as well as theoreticians new to safety-critical systems: software developers and architects, system engineers and operators of industrial plants, computer scientists and graduate students who know program analysis and temporal logic but wish to understand how safety concerns play out in practice, etc.

Resource Management in Cloud Computing

Cloud computing is an increasingly popular paradigm of offering services over the Internet; it is also an active area of research. One key area that calls for innovative approaches, is proper resource management. Considering that cloud systems are very large in size and business value (and will probably get only larger over time), and serve very large numbers of users, proper management of cloud resources is not a trivial concern. Resource management issues are not very simple, and improved solutions are likely to bring large financial and other benefits to both users and vendors of cloud systems.

This talk deals with the formulation and basic understanding of solutions to problems such as the following: How can cloud resources be allocated efficiently cloud vendors, given a multi-tiered business system on which user demands are time-varying? How can cloud services be metered in a time-varying way (rather than at a fixed rate that takes no account of changing conditions) to bring advantages to both cloud vendors and cloud users?

Sustainability in IT Systems

While we all appreciate the ways in which information technology (IT) has revolutionized society, it is becoming increasingly obvious that the full benefits of IT will be available to us and succeeding generations only if care is taken to use it responsibly. For instance, large data centers now consume power at the rate of megawatts, just like large industrial concerns. In the case of large systems, in addition to the power consumed by the computing devices themselves (all of which gets dissipated in the form of heat), the power consumed by the inevitable air-conditioning systems is also significant. Data centers and clouds now also have to deal with issues such as time-varying electricity tariffs, fluctuating power availability from renewable energy sources, and demand response. For such and other reasons as well, IT systems, large ones in particular, need to be designed to be more sustainable (in various ways), and concerns about effective resource usage are likely to play an important role in computing research, and in IT as well.

This talk presents a brief look at a few of the more important problems in sustainable computing and green IT, highlight some promising research directions that are being attempted, and also briefly touch upon some open problems and further issues that may be worth pursuing.