Service Differentiated Provisioning in Dynamic WDM Networks Based on Set-Up Delay Tolerance

Abstract

Optical networks are expected to provide a unified platform for a diverse set of emerging applications (three-dimensional TV, digital cinema, e-health, grid computing, etc). The service differentiation will be an essential feature of these networks. Considering the fact that users have different levels of patience for different network applications, referred to as set-up delay tolerance, it will be one of the key parameters for service differentiation. Service differentiation based on set-up delay tolerance will not only enable network users to select an appropriate service class (SC) in compliance with their requirements, but will also provide an opportunity to optimize the network resource provisioning by exploiting this information, resulting in an improvement in the overall performance. Improvement in network performance can be further enhanced by exploiting the connection holding-time awareness. However, when multiple classes of service with different set-up delay tolerances are competing for network resources, the connection requests belonging to SCs with higher set-up delay tolerance have better chances to grab the resources and leave less room for the others, resulting in degradation in the blocking performance of less patient customers. This study proposes different scheduling strategies for promoting the requests belonging to smaller set-up delay tolerance SCs, such as giving priority, reserving some fraction of available resources, and augmenting the research space by providing some extra paths. Extensive simulation results show that 1) priority in the rescheduling queue is not always sufficient for eradicating the degradation effect of high delay tolerant SCs on the provisioning rate of the most stringent SC, and 2) by utilizing the proposed strategies, resource efficiency and overall network blocking performance improve significantly in all SCs.

Delay tolerance aware queueing and scheduling for differentiated services in dynamic survivable WDM networks

“…They proposed different queuing policy algorithms for service differentiated provisioning with delay tolerance. The same authors, in [14], complete their work with detailed performance analysis.…”

“…To the best of our knowledge, only [14] investigates the issue of how the blocking performance of the SC with stringent delay tolerance would be affected by the improvement of the overall blocking probability. The study shows that the SC with smaller delay tolerance may suffer from more blocking probability.…”

“…For simplicity, it is assumed that the capacity of each fiber link is limited to 2 wavelengths. Furthermore, delay tolerance to holdingtime ratio ðt d =t h Þ for the three service classes are ð0:01; 0:5; 3Þ [14]. Whenever an incoming connection fails to establish its lightpath, the connection request is said to be blocked.…”

“…On the other hand, for non-real-time applications such as backup storage, e-science, and grid computing t d may extend to hours with similar values for the t h . For service differentiation with delay tolerance, it is proposed to use t d =t h as a parameter to differentiate the services [14]. A categorization of different applications for different service classes (SC) according to their t d to t h ratio are presented in [14].…”

“…By separating the queue of SC with small t d and applying the scheduling algorithm to it, we obtain better acceptance rate for this SC. In contrary to previous study [14], we use shared-path protection for provisioning without assuming precomputed k-shortest paths for routing.…”