Design And Implementation Of Geo-distributed Cloud Storage Middleware

Consumer cloud storage(CCS) provides geo-distributed storage service to personal users. In recent years, Dropbox, Google, Microsoft, and Baidu have been competing in the market of CCS services. While the comparative metric, storage capacity per user has outgrown the needs of most users. The next stage of this competitive landscape is taking shape, as service providers are seeking to differentiate themselves from the competition through third-party applications. Unfortunately, there is very little visibility into the performance of these API services, even though they are primary determinants of the end user experience on these storage applications.CcsMeasure is presented as a middleware for measurement and analysis of CCS APIs. The system consists of three layers. The top layer calls a specific API according to the command line arguments, and records the result into log files. The middle layer contains independent modules to implement the API protocols of CCSs. The interface between the top layer and the middle layer is predefined to contain the core set of CCS APIs. The bottom layer is an HTTP(s) client and an RTT measurement module built with Socket APIs and OpenSSL APIs. This layer records domain names and IPs used in each API call as well as the RTT for an in-depth understanding of the underlying system designs.The measurement system is deployed on 13 PlanetLab vantage points covering 9 countries and 5 continents, and periodically measures each API of the four CCSs during 20 consecutive days in April, 2015. The results reveal significant differences in terms of underlying system design, which result in performance divesity visible to the user through the storage application. Our research provides practical guidance for service providers to optimize their API performance, and for developers to improve the experience of third-party applications.