Towards the Description and Execution of Transitions in Networked Systems

Abstract

Today's distributed systems have to work in changing environments and under different working conditions. To provide high performance under these changing conditions, many distributed systems implement adaptive behavior. While simple adaptation through parameter tuning can only react to a limited range of conditions, a switch between different mechanisms at runtime enables broader adaptivity. However, distributed systems that switch mechanisms at runtime lack a clear abstraction for the adaptive behavior and, thus, usually interleave the adaptation and actual application logic. This leads to complex and error-prone systems that are hard to maintain and not easy to extend. In this paper, we analyze the adaptations of two distributed systems from different application domains. We identify recurring requirements as well as life cycles of transitions between mechanisms. Based on this, we present a framework that provides a clear abstraction of the underlying transition logic and manages the transitions at runtime. The concept is applied to the two example systems to practically evaluate its benefits. We show that our approach leads to less complex realizations of the adaptive behavior and allows new mechanisms to be integrated easily.

Analysing and leveraging client heterogeneity in swarming-based live streaming

“…Thus we both theoretically and practically show the potential of using a combination of primitive scheduling mechanisms to improve overall performance in mesh-/pull-based media streaming. These results are encouraging to consider using primitive scheduling mechanism combinations in mesh-based as well as hybrid streaming and enable seamless switching (transitions) between them as proposed in [6].…”

“…The results presented in this paper are encouraging in that SCHEDMIX could be used as an alternative to complex scheduling strategies in the growing number of scenarios where peer heterogeneity is inevitably given, e.g., when bandwidth-constrained mobile users meet well-connected and high-capacity home users. Besides, the results could be used in the planning of transitions [6] between strategies when environmental conditions change.…”

Tcep

“…The T transition engine coordinates how a transition is performed over the life cycle of a transition [15], i.e., from its invocation to its completion. The life cycle of a transition is de ned by the two transition strategies.…”

A Framework for Publish/Subscribe Protocol Transitions in Mobile Crowds

“…Each strategy consists of four basic steps: (i) start the new protocol through the lifecycle manager, (ii) transfer state to the new protocol, relying on the subscription storage, (iii) relay application and network interaction to the new protocol by utilizing the respective proxies, and (iv) terminate the previously running protocol. Before actually relaying all interaction to the new protocol, the strategy needs to wait until the lifecycle manager signals successful startup of the new protocol, as discussed in [9]. Depending on the protocol, startup might involve the formation of a topology and, as a consequence thereof, some message exchange unrelated to the application payload.…”

“…This allows us to adjust the performance vs. cost characteristics of the system based on the current conditions and application requirements, while at the same time relying on rather simple pub/sub protocols with well-known properties. Extending the concept of transitions and their corresponding lifecycles as proposed in [9], our framework addresses challenges related to the state transfer as well as message translation during the execution of a transition. We argue that existing protocols can be integrated easily, as they only need to provide access to stored subcriptions through a basic and easily extendable subscription model.…”