Towards checkpointing Grid architecture

Contemporary Grid environments are featured by an increasingly growing virtualization and distribution of resources. Such situations impose greater demands on load-balancing and fault-tolerant capabilities. The checkpoint-restart mechanism seems to be the most intuitive tool that can fulfill the specific requirements. One of the goals of the CoreGRID Network of Excellence is to define the high-level checkpoint-restart Grid Service and to locate it among other Grid Services. We aim to define both the abstract model of that service and the lower layer interface that will allow the service to cooperate with the diverse existing and future checkpoint-restart tools. The paper is the first step leading to achieving this goal. It includes the overall sketch of the architecture of the considered service and its connection with the actual checkpoint-restart tools. Additionally, the work on low-level checkpoint restart tools to be used in the "proof of concept" implementation and integration is mentioned.