Abstract

We present a new protocol for the distributed detection of garbage, suitable for a low-level distributed object-support system. It is based on realistic assumptions: messages may be lost or duplicated; sites may crash; an object may migrate or be deleted. The protocol uses only information local to each site, or exchanged between pairs of sites; no global mechanism is necessary. It is parallel and should scale to extremely large systems. It takes into account the object-finding protocol. Its interface is designed for maximum independence from other components.

Citations