Real-Time Guarantees in Flexible Advance Reservations

Abstract

Abstract-This paper deals with the problem of scheduling workflow applications with Quality of Service (QoS) constraints, comprising real-time and interactivity constraints, over a serviceoriented Grid network. A novel approach is proposed, in which high-level Advance Reservations, supporting flexible start and end time, are combined with low-level soft real-time scheduling, allowing for the concurrent deployment of multiple services on the same host while fulfilling their QoS requirements. By undertaking a stochastic approach, in which a-priori knowledge is leveraged about the probability of activation of the application workflows within the reserved time-frame, the proposed methodology allows for the achievement of various trade-offs between the need for respecting QoS constraints (user perspective) and the need for having good resource saturation levels (service provider perspective).

Advance reservations for distributed real-time workflows with probabilistic service guarantees

“…However, it is not necessary to actually consider each time-slot individually. In fact, it is possible to exploit a methodology for grouping advance reservations whose principles have already been introduced in [11]. Basically, the entire set of time-slots t k ∈ I is partitioned into G disjoint time-slices G {I h } h=1,...,G , of non-uniform duration of tn h time-slots, over which the set of applications does not change, formally: ∀t k ∈ I, t k ∈ I h =⇒ ∀t j ∈ I h , A(t j ) = A(t k ).…”

“…Based on the analysis in [11], the utilization assigned to the service τ (a) i of application a ∈ A can be considered as a discrete random variable u (a) i with a finite number of possible values, and with a probability distribution that changes for each time-slice I h ∈ G (a) .…”

“…2) Probability of conflicting sets of advance-reservations: For any given group I h ∈ G, that derives from the grouping process described in [11] and recalled in Section IV-A2, considering the arbitrariety (and independence) in the time instants in which users may request activations of the services, it is possible to compute the probability P j, B (I h ) of overlapping activation of the services in any possible subset B of A(I h ) on each given host j ∈ H, within any interval I h ∈ G :…”

Optimum allocation of distributed service workflows with probabilistic real-time guarantees

“…When evaluating the admission of a set of new applications A new , the time horizon of interest from the advance reservations viewpoint is the union I of the time-intervals of all the new applications to be admitted: I a∈A new I (a) . As shown in [15], the entire set of time-slots t k ∈ I may be logically partitioned into G disjoint time-slices G {I h } h=1,...,G , of nonuniform duration of tn h time-slots each, over which the set of potentially overlapping applications does not change. Formally:…”

“…Based on the analysis in [15], the utilization assigned to the service τ (a) i of application a ∈ A can be considered as a discrete random variable u (a) i with a finite number of possible values, and with a probability distribution that changes for each time-slice I h ∈ G (a) :…”