HydraNet-FT: network support for dependable servors

Abstract

This thesis proposes an infrastructure to dynamically replicate services across an internetwork and have them provide a single fault tolerant service access point to clients. For service replication, it relies on previous work for network support for large-scale service scaling within the HydraNet effort. The current work develops a means to combine service replication with TCP communication service to provide fault-tolerant services in a fully client-transparent fashion. The HydraNet-FT infrastructure consists of two components: host servers and redirecting. Host servers are hosts that are specially equipped to act as servers for replicated and fault-tolerant services. The location of the host servers is known to the redirecting, specially equipped routers that detect requests for replicated services and direct the requests to them. Host servers allow one-to-many message delivery from the client to servers and many-to-one message delivery from the servers to the clients. They also provide a low-latency failure estimator to determine server unavailability due to failure or congestion. A management protocol allows to dynamically install or remove service replica and reconfigure the system in case of failure or prolonged congestion of a server in the system. The thesis studies the results obtained by implementing the infrastructure on an experimental testbed. Throughput results indicate that the amount of overhead imposed by using the HydraNet-FT scheme is not unreasonably high.