Implementation of a Reliable Remote Memory Pager

Traditional operating systems use magnetic disks as paging devices, even though the cost of a disk transfer measured in processor cycles continues to increase. In this paper we explore the use of remote main memory for paging. We describe the design, implementation and evaluation of a pager that uses main memory of remote workstations as a faster-than-disk paging device and provides reliability in case of single workstation failures. Our pager has been implemented as a block device driver linked to the DEC OSF/1 operating system, without any modi cations to the kernel code. Using several test applications we measure the performance of remote memory paging over an Ethernet interconnection network and nd it to be faster than traditional disk paging. We evaluate the performance of various reliability policies and prove their feasibilityeven over low bandwidth networks, like Ethernet. We conclude that the bene ts of reliable remote memory paging in workstation clusters are signi cant today and will probably increase in the near future