Advancement Proposal Storage Embedded Networks (SEN) and Adaptive Caching using Multiple Experts

The gap between CPU speeds and the speed of the technologies providing the data is increasing. This causes the performance of processes to be limited by the performance of the storage devices, the networks and the buses. Furthermore, the number of CPUs that share these data access resources is growing exponentially. Caching, prefetching and parallelism are some of the techniques used today to cope with I/O latency and system scalability to support more users. This paper describes the two major contributions of our ongoing research on distributed data access. The first contribution is the design of the Storage Embedded Networks (SEN) architecture that aims to improve user response times and scalability on the Internet by better distribution of caches. SEN architecture is composed of trusted routers embedded with volatile and non–volatile storage that snoop bypassing objects for caching. Requests are checked by every hop, thus ensuring the transmission of the closest copy on the data path and load reduction at the upstream. The two main control overheads of other architectures, connection establishment and continuous cache communications, do not exist in SEN. The second contribution is the design of adaptive caching schemes using multiple ex