Co-allocation of compute and network resources using HARC
Jon MacLaren · 2007
HARC-the Highly-Available Resource Co-allocator-is a system for reserving multiple resources in a coordinated fashion.HARC can handle multiple types of resource, and has been used to reserve time on supercomputers distributed across a nationwide testbed in the United States, together with dedicated lightpaths connecting the machines.HARC makes these multiple allocations in a single atomic step; if any resource is not available as required, then nothing is reserved.To achieve this "all or nothing" behavior, HARC treats the allocation process as a Transaction, and uses a phased commit protocol.The Paxos Commit protocol to ensure that there is no single point of failure in the system, which, if correctly deployed, has a very long Mean Time To Failure.Here we give an overview of HARC, and explain how the current HARC Network Resource Manager (NRM) works, and is able to set-up and tear-down dedicated lightpaths. Lighting the Blue Touchpaper for UK e-Science -Closing Conference of ESLEA Project March 26-28