PropScale: An update propagator for joint scalable storage

Abstract

In the era of Web 2.0 and the apparent dawn of Web 3.0 web pages are dynamic and personalized. As the result, the load of web servers rapidly increases. Moreover, the upcoming load boost is impossible to predict. Although deceptively funny, the term of success - tolerant architectures has been coined. A number of web services actually failed because of their initial success. In order to achieve success-tolerance the server architectures must be scalable. Nowadays almost all components of systems can certainly be multiplied. The only exception is the storage constituent. The usual solution with one strong relational database is unsatisfactory. Thus, designers introduce additional (NO)SQL storage facilities. From this point one has a number of separate data sources that can apparently get inconsistent with each other. Special software must be developed to synchronize them. This means more bugs to fix, more code to maintain and more money to spend. In this paper we present a new technique to introduce a number of non-homogenous storage units into a system. The solution consists of an algorithm that propagates updates among disparate (NO)SQL storages built into a system. © 2013 Springer-Verlag.