Researchers: Charles Zhang, Jason Leigh, Luc Renambot, Venkatram Vishwanath
LambdaRAM (Optically Connected Memory Cache) is an application being developed to address long-haul latency in optical networks.
LambdaRAM is a tool that aggregates pools of memory in clusters of gigabit-connected computers to provide a massive data cache. This minimizes access latency when data-intensive applications need to fetch data from distantly located data stores.
LambdaRAM utilizes available network bandwidth to aggressively prefetch data before an application needs it. Applications access LambdaRAM via an API that provides a single contiguous memory image. LambdaRAM is currently used in TeraScope (EVL’s visual data mining software) to provide data correlation algorithms with fast access to distributed database tables.
Date: August 1, 2001 - Ongoing