47 ;;Quote: halting failures of distributed computation when process halts without incorrect actions

49 ;;Quote: assume that networks are hierarchical without internal partitioning

49 ;;Quote: logical failure handling: run a protocol to agree on ordering of failure event relative to other events

51 ;;Quote: GBCAST for group broadcast of failures, recoveries, etc; can locate most up-to-date information

52 ;;Quote: ABCAST for when all sites need data in the same order

52 ;;Quote: CBCAST for casual broadcast to maintain a partial ordering system wide

56 ;;Quote: checkpoints by first broadcasting GBCAST to create a logical time

59 ;;Quote: site view of all operational sites with their incarnation numbers

61 ;;Quote: implementation of ABCAST, CBCAST, GBCAST with delivery queue

72 ;;Quote: ISIS defines a resilient object as a coordinator with cohorts for passive backup

74 ;;Quote: ISIS primitives allow good performance while tolerating failure; simplify design of distributed systems