”
Esta palestra é co-organizada pelo ACM e pelo Grupo de Sistemas Distribuídos do DI.
“In this talk I will present Sprint, a middleware infrastructure designed at the University of Lugano (USI), Switzerland, targeting high performance and high availability data management. Sprint extends the functionality of a standalone in-memory database to a cluster of commodity shared-nothing servers. The execution model is very simple: Transaction synchronization and commitment relies on a very efficient total-order multicast.
Differently from several previous approaches, Sprint does not require accurate failure detection to ensure strong consistency (i.e., serializability), providing very fast reaction to failures. Besides discussing Sprint’s design and algorithms, I will present some performance results.”
“