Loading and Querying Data on Distributed Virtualized Web Application Servers
Moritz Mack · KTH Publication Database DiVA (KTH Royal Institute of Technology) · 2009
Virtualized web application servers within computational clouds, such as the GoogleApp Engine, generally restrict resource usage and therefore provide limited,relationally none-complete query facilities only. This work investigates how scalable,reliable and a more powerful access to the App Engine Datastore can beaccomplished and an Optimized Distributed Datastore Exchange (ODDSE) ispresented. Being aware of the App Engine’s resource restrictions ODDSE provides areliable and failure safe query interface to transparently exchange data with thedistributed Datastore using full SQL or AmosQL. ODDSE therefore wraps Datastorerelations and utilizes the extensible database system Amos II to compensate formissing query facilities in Google’s relationally none-complete query language GQL.Under the covers ODDSE furthermore implements an adaptive and reliablemanagement of interactions with App Engine servers. For scalability and highperformance the interaction is parallelized and continuously adapted. Theperformance of ODDSE is evaluated and compared to a similar system showing itsconsiderably good results for both bulk uploading and query processing.