Towards autonomic workflow management systems

Abstract

In a world of dynamic and discontinuous change, systems constantly need to adapt to new conditions so that they can survive and flourish in their environment. Autonomic computing emerged as a research field that takes up this challenge and aims to build systems that are capable of adapting automatically to dynamically changing environments (Self-configuring), discovering, diagnosing and reacting to disruptions (Self-healing), monitoring and tuning resources automatically (Self-optimizing) and anticipating, detecting, identifying and protecting themselves from attacks (Self-protecting) [3]. A major application area for autonomic computing is intended to be system administration, aiming to free system administrators from the details of system operation and maintenance [8], improving robustness of systems and decreasing total cost of ownership. However, the vision of autonomic computing does not need to be restricted to the area of system administration. For example, much research has been done in the area of process-aware information systems [2] such as Workflow Management, Enterprise Resource Planning, Business-to-Business and Customer Relationship systems to effectively and efficiently deal with change on different levels and scales. Frequent questions in these domains include: How can changes to workflows be accommodated? How can flexibility and adaptability of running workflow instances be increased? How can workflow management systems themselves optimize workflow definitions? The type of questions raised here seems to address issues that are similarly addressed by research in autonomic computing, where dealing with change represents a major concern. However, little research has been done on the intersection between these two domains [4]. Based on this observation, this contribution aims to tackle the question: "Can the principles of autonomic computing be applied to workflow management - and if so, how?"

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 9 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.