DOLAR: Virtualizing Information Spaces to Support Information Expansion
Speaker: Kostas Saidis
University: Univ. of Athens
Room : A56
Abstract: The amount of information produced as well as consumed in the world is constantly expanding. Clearly, applications need to continuously scale-up to support increasing volumes of data. In addition, information expansion imposes a constant demand for supporting newly introduced and often unforeseen types of content. To cope with such a demand efficiently, existing services should operate atop constantly expanding information spaces, supporting diverse and heterogeneous items without expensive and time-consuming source-code revisions. To address this need, we introduce DOLAR, a service-neutral virtual information space framework which offers return of investment on existing services. DOLAR automates the introduction of new business-logic objects in terms of virtual ``content objects''. User-specified virtual objects are connected to storage artifacts and help applications realize uniform ``store-to-user'' information flows atop heterogeneous sources, while offering the reverse ``user-to-store'' flows with identical effectiveness and ease of use. DOLAR offers mechanisms to gradually support new types of storage choices, while virtual objects help business-logic essentially operate in isolation of any low-level information space idiosyncrasies. The combined effect is that DOLAR enables applications to automatically include newly introduced types of items in their existing service provision. We show how DOLAR has assisted in the cost-effective gradual expansion of a production-quality digital library, while experimentation shows that our approach imposes minimal overheads and DOLAR-based applications scale as well as any underlying datastore(s).