الفهرس 
Handling cloud paaS platforms heterogeneity
AbstractOnly 14 pages are availabe for public view
 |  | from | 138 |  |  |
| | | from | 138 | | |
Abstract
The providers of Platform as a Service (PaaS) have their own specific APIs (i.e., heterogeneous). These specific APIs make developers locked inside a specific platform and they can’t able to port their applications easily among di erent platforms. As a result, vendor lock-in problem has taken place. One solution for this problem is to use generic APIs with speci c adapters for di erent PaaS platforms to implement portable applications. However, any update in a PaaS speci c-API makes its corresponding adapter to become unusable, which causes what is called API dynamic adaptation prob- lem. Therefore, the work in this thesis provides three main contributions to solve these problems, (1) extend the Compatible One Application and Platform Service (COAPS) generic deployment-API, (2) propose service-based generic-APIs (Std-PaaS APIs) and (3) propose a Semantic-based Generation of Generic-API adapters (STAGER) framework. According to the rst contribution, the COAPS generic deployment-API has been extended to deploy applications on Google App Engine (GAE) platform, besides Cloud- Foundry (CF) and OpenShift (OS). Regarding the second contribution, the proposed service-based generic-APIs (called Std-PaaS APIs) can be used by cloud developers to implement generic applications. In addition, the proposed Std-PaaS APIs can include a set of generic APIs for each PaaS service (e.g., blob storage, datastore, messaging, etc.). Currently, the Std-PaaS APIs include two generic APIs; blob storage and NoSQL datastore services. With respect to the third contribution, the proposed STAGER framework is used to semi-automatically generate the adapters for any semantically annotated generic APIs. STAGER framework has been evaluated using the Std-PaaS APIs by generating their adapters for two PaaS platforms; Google App Engine (GAE) and Windows Azure