An Evaluation Model for GI SMEs migrating to geocloud services

Abstract

Summary The current thesis follows the ‘Design Science Research Methodology’ (DSRM) guidelines for the academic assessment of new IT innovations presented by Hevner et al. (2004). This work will contain six chapters: 1) Introduction to Cloud Computing, 2) Problem Context ,3) Problem Definition, 4) Solution Design and Development, 5) Demonstration, 6) Evaluation. The key findings are summarised below.

Background: The cloud is a new paradigm in IT and data management. Cloud technologies allow for data services (e.g. storage and processing) to be carried out using remote IT infrastructures (e.g. servers, software) via an internet connection. The cloud industry is growing rapidly and spatial or “geocloud” infrastructures are now an emerging niche within that industry.

Problem Context: This research investigates geocloud migration in Geographic Information Small to Medium Enterprises (GI SMEs). Previous research has suggested that 52% of SMEs do not engage in migration planning and that the evaluation methods adopted by SMEs tend not to be rigorous due to limited organisational resources (e.g. finance, time, IT skills). Furthermore, no cloud evaluation model has yet been developed which takes infrastructural, spatial data considerations into account in cloud environments. The aim of this research is to design a geocloud evaluation model, which can be easily applied by GI SMEs.

Solution Design and Development: A combined application index (Apdex) and Analytical Hierarchy Process (AHP) approach was designed in the current thesis which aimed to integrate both user needs and vendor testing. This approach involved reviewing relevant features across (geo) cloud platforms within a cloud matrix. GI SME decision makers were then presented with 81 feature scenarios and asked to rate their favourability relative to Apdex. Apdex calculations were then carried out for eight cloud attributes, rating their preferability between 0 and 1. In the case of variable cloud features (e.g. query response time), case specific performance testing was carried out to provide data for calculations. An overall cloud platform or scenario rating was then provided by weighing Apdex scores relative to cloud attribute importance, which had been adjusted relative to decision maker’s certainty.

Demonstration: The applicability of the proposed geocloud evaluation approach was demonstrated in three GI SMEs in this thesis. All enterprises had an interest in migrating applications or spatial data products to the cloud. Participating organisations were all different sizes and offer a good cross-section of GI SMEs as a whole. Following the methodology above, the favourability of different geo-SaaS platforms or I/PaaS scenarios was quantified relative to user needs. Analysed as a percentage, geo-SaaS evaluation had a wider spread in terms of favourability, whereas, I/PaaS scenarios were quite homogenous in terms of their favourability.

Evaluation: The proposed evaluation model, in its current form, is more suitable to evaluating geo-SaaS offerings than more complex I/PaaS scenarios. No significant differences were observed when quantifying the favourability of different I/PaaS scenarios, in two of three cases. This result is unrealistic and likely occurred because feature preference questions were overly focussed on geo-SaaS. Withstanding this content-related shortcoming, the overarching framework presented here is still likely to be use-worthy, if improved. Each element of the proposed approach was evaluated with feedback from the participating decision makers,

8 | P a g e Apdex: The applicability of Apdex to geocloud evaluation was rated with moderate positivity by decision makers, who stated that it was a straightforward method with which cloud favourability could be quantified.

AHP: Decision makers highly rated the AHP element of the current research as it simplified the complexity of cloud selection.

The need for further validation of this methodology is also highlighted. Ideally, a longitudinal analysis of outcome satisfaction, post cloud migration, would be carried out to assess the current model. Future work on this evaluation model should focus on the standardisation of evaluation matrices or indices and the quantification of the performance of different spatial cloud infrastructures such as PostGIS and Oracle Spatial.