Providing QoS strategies and cloud‐integration to web servers by means of aspects

Abstract

SUMMARYThe main responsibilities of a web server are to listen from the communication channel and to prepare replies to requests. Additional responsibilities include adapting processing activities, for example, through scheduling or request filtering, so as to satisfy Quality of Service (QoS) requirements. Typical QoS-related concerns address behavioural constraints (e.g. response time bounds, satisfiable by scheduling the most urgent requests first) and resource monitoring, for optimal use. Although such concerns are spread across several web server components, they should be handled separately from communication-related ones, for the sake of modularity.For this purpose, we advocate recourse to aspect-oriented programming and illustrate it by showing how a QoS-related layer can be smoothly superimposed on top of a well-known, unmodified, web server Jigsaw. As part of the provided support, requests are assessed and partitioned into those that can be appropriately handled by using local resources and those needing further resources. For the latter requests, cloud-based resources are gathered. Aspect-orientation enables new QoS-related code to be separated from web server modules so as to keep existing code unaltered, while runtime behaviour is modified with the measures needed to handle QoS concerns.

A Hybrid Neuro–Wavelet Predictor for QoS Control and Stability

“…As can be seen, the neural network manages to closely predict even relatively small variations of the trends. The output of the neural network was then given to a resource management service to perform allocation requests in terms of needed bandwidth and virtual machines [12].…”

“…Moreover, the request for stable and continuous internet services has reached a pressing priority. Generally, the solutions employed by service providers to adapt server-side resources on-the-fly have been based on content adaptation and differentiated service strategies [1,9,3,12,13]. However, such strategies could over-deteriorate the service during load peaks.…”

Improving Files Availability for Bittorrent Using a Diffusion Model

“…Index k is used in equation (10) to refer to file fragment z i k . Indeed, since nodes need and offer their own file fragments, the ordered set of nodes referred by a given node depend on resource z k , i.e.…”

“…Therefore, the proposed work aims at supporting Quality of Service and dependability of P2P systems by attributing a priority on both the fragment to spread and the destination peer. This in turn increases availability and performances [9], [10], [11], as well as consistency [1].…”