For the third edition of the INESC-ID Distinguished Lecture series, we will welcome Robbert van Rennesse from the Cornell University for a talk titled “Chain Replication Revisited”.

Date & Time: 17 November, 12h00

Location: Anfiteatro Abreu Faro at Instituto Superior Técnico

Abstract: “Chain Replication tolerates f crash failures with only f+1 replicas, half of what Paxos needs, and is used in production storage systems such as EBS, Delta and RADOS. This talk revisits Chain Replication twenty years after its introduction. Quorum protocols such as Paxos never need to tell a slow process from a crashed one. Our new protocol, Chain Replication Revisited (CR2), takes the opposite approach: it makes slowness indistinguishable from a crash. Replicas hold leases from a small set of lessors and stop themselves when they can no longer renew them. Lessors keep no persistent state and never communicate, and the only assumption is a bound on clock drift. With failure detection that is guaranteed accurate, the chain can manage its own membership without a consensus service. CR2 also persists updates in parallel rather than one replica at a time and can recover even when every replica crashes (assuming stable storage).”

Bio: “Robbert van Renesse is a Professor of Computer Science at Cornell University. His research focuses on fault-tolerant and scalable distributed systems. He co-designed chain replication and has worked on systems ranging from Amoeba to Ziplog. He has also co-founded several companies, including Reliable Network Solutions and Exostellar, bringing systems research into practice. He received his Ph.D. from the Vrije Universiteit Amsterdam under Andrew S. Tanenbaum. He is an ACM Fellow and co-editor-in-chief of ACM Transactions on Computer Systems.“