Use this URL to cite or link to this record in EThOS:
Title: Predictive dynamic resource allocation for web hosting environments
Author: Al Ghamdi, Mohammed A.
ISNI:       0000 0004 2726 4735
Awarding Body: University of Warwick
Current Institution: University of Warwick
Date of Award: 2012
Availability of Full Text:
Access from EThOS:
Access from Institution:
E-Business applications are subject to significant variations in workload and this can cause exceptionally long response times for users, the timing out of client requests and/or the dropping of connections. One solution is to host these applications in virtualised server pools, and to dynamically reassign compute servers between pools to meet the demands on the hosted applications. Switching servers between pools is not without cost, and this must therefore be weighed against possible system gain. This work is concerned with dynamic resource allocation for multi-tiered, clusterbased web hosting environments. Dynamic resource allocation is reactive, that is, when overloading occurs in one resource pool, servers are moved from another (quieter) pool to meet this demand. Switching servers comes with some overhead, so it is important to weigh up the costs of the switch against possible system gains. In this thesis we combine the reactive behaviour of two server switching policies – the Proportional Switching Policy (PSP) and the Bottleneck Aware Switching Policy (BSP) – with the proactive properties of several workload forecasting models. We evaluate the behaviour of the two switching policies and compare them against static resource allocation under a range of reallocation intervals (the time it takes to switch a server from one resource pool to another) and observe that larger reallocation intervals have a negative impact on revenue. We also construct model- and simulation-based environments in which the combination of workload prediction and dynamic server switching can be explored. Several different (but common) predictors – Last Observation (LO), Simple Average (SA), Sample Moving Average (SMA) and Exponential Moving Average (EMA), Low Pass Filter (LPF), and an AutoRegressive Integrated Moving Average (ARIMA) – have been applied alongside the switching policies. As each of the forecasting schemes has its own bias, we also develop a number of meta-forecasting algorithms – the Active Window Model (AWM), the Voting Model (VM), the Selective Model (SM), the Dynamic Active Window Model (DAWM), and a method based on Workload Pattern Analysis (WPA). The schemes are tested with real-world workload traces from several sources to ensure consistent and improved results. We also investigate the effectiveness of these schemes on workloads containing extreme events (e.g. flash crowds). The results show that workload forecasting can be very effective when applied alongside dynamic resource allocation strategies.
Supervisor: Not available Sponsor: Engineering and Physical Sciences Research Council (EPSRC) (EP/C538277/1)
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID:  DOI: Not available
Keywords: QA76 Electronic computers. Computer science. Computer software