Summary Experience has shown that a priori created static resource allocation plans are vulnerable to runtime deviations and hence often become uneconomic or highly exceed a predefined soft deadline. The assumption of constant task execution times during allocation planning is even more unlikely in a cloud environment where virtualized resources vary in performance. Revising the initially created resource allocation plan at runtime allows the scheduler to react on deviations between planning and execution. Such an adaptive rescheduling of a many‐task application workflow is only feasible, when the planning time can be handled efficiently at runtime. In this paper, we present the static low‐complexity resource allocation planning algorithm (LCP) applicable to efficiently schedule many‐task scientific application workflows on cloud resources of different capabilities. The benefits of the presented algorithm are benchmarked against alternative approaches. The benchmark results show that LCP is not only able to compete against higher complexity algorithms in terms of planned costs and planned makespan but also outperforms them significantly by magnitudes of 2 to 160 in terms of required planning time. Hence, LCP is superior in terms of practical usability where low planning time is essential such as in our targeted online rescheduling scenario.

Grid computing requires network services beyond what is currently provided by the Best-Effort Internet. Among the different approaches towards network Quality of Service, aggregate scheduling, which maps micro-flows to a small number of different service classes, offers sufficient scalability up to the size of the Internet. The Differentiated Services architecture of the Internet Engineering Task Force is such an approach. Recently further Best-Effort like services have been proposed that are based on aggregate scheduling, like the less than Best-Effort Scavenger Service and the Alternative Best-Effort Service. In this paper we introduce the existing aggregate based approaches to Quality of Service. We take a heterogeneous mix of Transmission Control Protocol and User Datagram Protocol flows into account and complement services by adding transport protocol specific elements, if appropriate. In particular we show multi-class applications that are designed to apply a Guaranteed Rate Service and a Scavenger Service in parallel. Doing so we can show a performance gain and achieve a more economical use of the available resources without impacting responsive Best-Effort flows.

In this paper we describe the implementation of a network providing advanced services such as a Premium service that aims at providing low loss, low delay, and low delay jitter and an Olympic service that allows for a service differentiation in terms of delay within three additional classes. Our implementation of this network is based on the Differentiated Services Architecture, which is the most recent approach of the Internet Engineering Task Force towards quality of service. Access to service classes is controlled by a bandwidth broker, which can perform traffic engineering by means of multiprotocol label switching. The Premium service is implemented as expedited forwarding and the Olympic service as a group of assured forwarding per-hop-behavior. We present a thorough evaluation of the proposed services implemented by the careful assignment of micro-flows to a small set of per-hop behaviors.

High-end networked applications such as distance visualization, distributed data analysis, and advanced collaborative environments have demanding quality of service (QoS) requirements. We focus on making policy decisions when users attempt to make reservations for network bandwidth across several administrative network domains that are controlled by a bandwidth broker. We present a signalling protocol that facilitates the establishment of a distributed policy decision point as well as the establishment of a direct signalling channel between the source and end domains.

Parallel programmers typically assume that all resources required for a program’s execution are dedicated to that purpose. However, in local and wide area networks, contention for shared networks, CPUs, and I/O systems can result in significant variations in availability, with consequent adverse effects on overall performance. We describe a new message-passing architecture, MPICH-GQ, that uses quality of service (QoS) mechanisms to manage contention and hence improve performance of message passing interface (MPI) applications. MPICH-GQ combines new QoS specification, traffic shaping, QoS reservation, and QoS implementation techniques to deliver QoS capabilities to the high-bandwidth bursty flows, complex structures, and reliable protocols used in high-performance applications-characteristics very different from the low-bandwidth, constant bit-rate media flows and unreliable protocols for which QoS mechanisms were designed. Results obtained on a differentiated services testbed demonstrate our ability to maintain application performance in the face of heavy network contention.