Amadeus S.a.s.

Wayne Rubenstein

Rudy Daniello

Luc Isnardy

Benoit Janin

Claudine Reynaud

Rémy Gole

Frédérick Ros

Remy Gole

Frederick Ros

A method and a system for maintaining coherence of cache contents in a multi-tiered architecture of servers are described. This includes a front tier of satellite servers, each operating a local cache, and a middle tier of central servers each operating a central cache. Central servers interface with databases through database servers to retrieve the data elements used to construct objects and store them in central caches. Once constructed, objects are attributed a time-to-live (TTL) and stored in central caches then, forwarded to the satellite servers where they are stored in local caches before being delivered to the software applications that have requested them. They are invalidated when outdated and reconstructed from a central server from where they are forwarded to all central caches and to the local caches where they are needed.

System and method to maintain coherence of cache contents in a multi-tier system aimed at interfacing large databases

The invention concerns a method and a system for maintaining consistency of a cache memory, accessible by multiple independent processes. The processes can share common data. The processes perform simultaneous data searching operations optionally followed by providing the data to the processes, a removal of same or an insertion of new data. The searching, removal and insertion operations, are comprehensively executed once they have been initiated by the independent processes. They are executed excluding one another when they must operate on common data. The removal or insertion operations are each completely reversible. In that context, the invention provides that the operations for providing, removing or inserting the data have a finite or bound duration of execution so as to prevent any locking. In particular, the insertion operation, which includes a step of releasing an input when a partition of the cache memory is full, if it requires browsing through the entire partition to make a selection, is however possibly interrupted if the maximum browsing time is reached. In that case, the most appropriate input, among the already browsed inputs, is released. The inputs of said partition are browsed exhaustively, which results from storing the last consulted address of the partition after each release.

Method and system for maintaining consistency of a cache memory accessible by multiple independent processes

Growing community of inventors

Nice, France

Wayne Rubenstein