PRACTI replication for Large-Scale Systems

M. Dahlin, L. Gao, A. Nayate, A. Venkataramani, P. Yalagandula, J. Zheng

We present PRACTI, a new approach and architecture for large-scale replication. PRACTI systems can replicate or cache any data on any node (Partial Replication), provide a broad range of consistency and coherence guarantees (Arbitrary Consistency), and permit any node to share updates with any other node (Topology Independence). Our PRACTI architecture yields two significant advantages. First, it provides better trade-offs than existing mechanisms: our prototype dominates existing approaches by providing as much as an order of magnitude better bandwidth and storage efficiency than AC-TI replicated server systems, as much as an order of magnitude better synchronization delay compared to PR-AC hierarchical systems, and consistency guarantees not achievable by PR-TI per-object replication systems. Second, our architecture's flexibility simplifies the design of replication systems by allowing a single framework to subsume a broad range of existing systems and to reduce development costs for new ones. For example, we use our PRACTI prototype both to emulate existing server replication, client-server, and object replication systems and to implement novel policies that improve performance for mobile users, web edge servers, and grid computing.

Full paper
Extended version paper