Ultra-Large-Scale Systems

Ultra-Large Scale Systems - Systems of the future will push far beyond the size of today’s systems and systems of systems by every measure: number of lines of code; number of people employing the system for different purposes; amount of data stored, accessed, manipulated, and refined; number of connections and interdependencies among software components; and number of hardware elements. They will be [http://www.sei.cmu.edu/uls/glossary.html#u ultra-large-scale (ULS) systems] .

The characteristics of ULS systems are already beginning to emerge in some of today’s systems, such as the global Internet or the Google search engine, but in the future, these characteristics will be ubiquitous and will challenge all of the assumptions of traditional computer science, software engineering, and systems engineering.

The sheer scale of ULS systems will change everything. ULS systems will necessarily be decentralized in a variety of ways, developed and used by a wide variety of stakeholders with conflicting needs, evolving continuously, and constructed from heterogeneous parts. People will not just be users of a ULS system; they will be elements of the system. Software and hardware failures will be the norm rather than the exception. The acquisition of a ULS system will be simultaneous with its operation and will require new methods for control. In ULS systems, these characteristics will dominate. Consequently, ULS systems will place unprecedented demands on software acquisition, production, deployment, management, documentation, usage, and evolution practices.

Software is the least well understood and the most problematic element of our largest systems today. Our current understanding of software and our software-development practices will not meet the demands of the future. To make significant progress in the size and complexity of systems that can be built and deployed successfully, we require a culture shift.

Initial ULS Systems Study

In 2006-2007, the Carnegie Mellon Software Engineering Institute, a federally funded research and development center, conducted a study on behalf of the U.S. Office of the Assistant Secretary of the Army (Acquisition, Logistics, & Technology). The study brought together experts from within and outside the field of software engineering and from a variety of institutions and organizations to consider the challenges of ultra-large-scale (ULS) systems.

"Ultra-Large-Scale Systems: The Software Challenge of the Future" (ISBN 0-9786956-0-7) is the product of this 12-month study. [http://www.sei.cmu.edu/uls/files/ULS_Book2006.pdf The report] , details a broad, multi-disciplinary research agenda for developing the ultra-large-scale systems of the future.

Research Agenda

Fundamental gaps in our current understanding of software and software development at the scale of ULS systems present profound impediments to the technically and economically effective achievement of significant gains in core system functionality. These gaps are strategic, not tactical. They are unlikely to be addressed adequately by incremental research within established categories. Rather, we require a broad new conception of both the nature of such systems and new ideas for how to develop them. We will need to look at them differently, not just as systems or systems of systems, but as socio-technical ecosystems. We will face fundamental challenges in the design and evolution, orchestration and control, and monitoring and assessment of ULS systems. These challenges require breakthrough research.

In its study, the [http://www.sei.cmu.edu/uls/the_report.html SEI author team] proposed a ULS systems research agenda for an interdisciplinary portfolio of research in at least the following areas:

[http://www.sei.cmu.edu/uls/human_interaction.html Human Interaction] - People are key participants in ULS systems. Many problems in complex systems today stem from failures at the individual and organizational level.Understanding ULS system behavior will depend on the view that humans are elements of a socially constituted computational process. This research involves anthropologists, sociologists, and social scientists conducting detailed socio-technical analyses of user interactions in the field, with the goal of understanding how to construct and evolve such socio-technical systems effectively.

[http://www.sei.cmu.edu/uls/computational_emergence.html Computational Emergence] - ULS systems must satisfy the needs of participants at multiple levels of an organization. These participants will often behave opportunistically to meet their own objectives.Some aspects of ULS systems will be “programmed” by properly incentivizing and constraining behavior rather than by explicitly prescribing. This research area explores the use of methods and tools based on economics and game theory (e.g., mechanism design) to ensure globally optimal ULS system behavior by exploiting the strategic self-interests of the system’s constituencies. This research area also includes exploring metaheuristics and digital evolution to augment the cognitive limits of human designers, so they can manage ongoing ULS system adaptation more effectively.

[http://www.sei.cmu.edu/uls/design.html Design] - Current design theory, methods, notations, tools, and practices and the acquisition methods that support them are inadequate to design ULS systems effectively.This research area broadens the traditional technology-centric definition of design to include people and organizations; social, cognitive, and economic considerations; and design structures such as design rules and government policies. It involves research in support of designing ULS systems from all of these points of view and at many levels of abstraction, from the hardware to the software to the people and organizations in which they work.

[http://www.sei.cmu.edu/uls/computational_engineering.html Computational Engineering] - New approaches will be required to enable intellectual control at an entirely new level of scope and scale for system analysis, design, and operation.ULS systems will be defined in many languages, each with its own abstractions and semantic structures. This research area focuses on evolving the expressiveness of representations to accommodate this semantic diversity. Because the complexity of ULS systems will challenge human comprehension, this area also focuses on providing automated support for computing the behavior of components and their compositions in systems and for maintaining desired properties as ULS systems evolve.

[http://www.sei.cmu.edu/uls/adaptive_system_infrastructure.html Adaptive System Infrastructure] - ULS systems require an infrastructure that permits organizations in distributed locations to work in parallel to develop, select, deploy, and evolve system components.This research area investigates integrated development environments and runtime platforms that support the decentralized nature of ULS systems. This research also focuses on technologies, methods, and theories that will enable ULS systems to be developed in their deployment environments.

[http://www.sei.cmu.edu/uls/system_quality.html Adaptable and Predictable System Quality] - ULS systems will be long running and must operate robustly in environments fraught with failures, overloads, and attacks. These systems must maintain robustness in the presence of adaptations that are not centrally controlled or authorized.

Managing traditional qualities such as security, performance, reliability, and usability is necessary but not sufficient to meet the challenges of ULS systems. This research area focuses on how to maintain quality in a ULS system in the face of continuous change, ongoing failures, and attacks. It also includes identifying, predicting, and controlling new indicators of system health (akin to the U. S. gross domestic product) that are needed because of the scale of ULS systems.

[http://www.sei.cmu.edu/uls/policy.html Policy, Acquisition, and Management] - Policy and management frameworks for ULS systems must address organizational, technical, and operational policies at all levels. Rules and policies must be developed and automated to enable fast and effective local action while preserving global capabilities. This research area focuses on transforming acquisition policies and processes to accommodate the rapid and continuous evolution of ULS systems by treating suppliers and supply chains as intrinsic and essential components of a ULS system.

The proposed research does not supplant current, important software research but rather significantly expands its horizons. Moreover, because it is focused on systems of the future, the SEI team purposely avoided couching descriptions in terms of today’s technology. The envisioned outcome of the proposed research is a spectrum of technologies and methods for developing these systems of the future, with national-security, economic, and societal benefits that extend far beyond ULS systems themselves.

External Links

[http://www.sei.cmu.edu/uls/ ULS Systems] - Carnegie Mellon® Software Engineering Institute's program for Ultra Large Scale Systems

[http://ulssis.cs.virginia.edu/ ULSSIS] - The Center for Ultra-Large Scale Software-Intensive Systems; the acronym is pronounced like "Ulysses"

[http://www.sei.cmu.edu/uls/files/ULS_Book2006.pdf Ultra-Large-Scale Systems: The Software Challenge of the Future] - The 2006 report for a 12-month study of ultra-large-scale systems software, sponsored by the United States Department of Defense

[http://www.sei.cmu.edu/uls/glossary.html ULS Systems Glossary]

[http://dsonline.computer.org/portal/site/dsonline/menuitem.9ed3d9924aeb0dcd82ccc6716bbe36ec/index.jsp?&pName=dso_level1&path=dsonline/2007/10&file=ox003news.xml&xsl=article.xsl Stepping Up to Long-Term Research] - IEEE Distributed Systems Online article on Ultra-Large Systems research

[http://lambda-the-ultimate.org/node/2874 Why Multi-Core is Easy and Internet is Hard] - a paper (and discussion) that touches on topics important in ULS research

[http://www.agorics.com/Library/agoricpapers.html The Agoric Papers] - Three papers on capability-based market-oriented computing (concepts that are the subject of some ULS Systems research), written by Mark S. Miller and K. Eric Drexler