Skip to Main Content
Distributed, event based applications executing on publish-subscribe middleware (a.k.a. event based middleware or EBM) are subject to many disruptive changes at runtime. These changes can be caused either by a change in environmental conditions such as the failure of an event broker node or can be caused by dynamically evolving application needs such as the sudden need to route events securely from a specific publisher to a set of subscribers. Ideally, the middleware should facilitate uninterrupted execution of applications without costly redeployment in the face of these changes. We formally study the requirements of such an "adaptive" middleware infrastructure and analyze adaptation in existing efforts. An adaptive overlay for a core level adaptive middleware with a chordal reliability ring is proposed on the basis of the study. We illustrate and prove that two of the adaptation triggers -node and link failures can be handled by the middleware in O(log n) time with the help of the chordal reliability ring.