University of Cambridge > > Computer Laboratory Systems Research Group Seminar > Ios: Why work when you can delegate?

Ios: Why work when you can delegate?

Add to your list(s) Download to your calendar using vCal

If you have a question about this talk, please contact Liang Wang.

The common wisdom when it comes to distributed consensus is to select one node as a leader (aka master, primary or distinguished proposer) who coordinates the system and the rest of the nodes (aka backups, replicas or followers) do as instructed by the leader. This centralised approach simplifies the tricky problem of consensus but places all the responsibility onto the leader, making them the bottleneck for system. Adding nodes to the systems decreases throughput, seriously limiting the scalability of systems built upon distributed consensus.

Ios is a new consensus algorithm which enables leaders to safely and dynamically delegate the responsibilities of consensus onto other nodes. The system is thus able handle higher workloads and is more scalability then traditional consensus algorithms. Ios is a work-in-progress and not production ready. If you’ve interested in learning more chat with me or take a look at the links below:

- – open source implementation of Ios, written in go

- – interactive simulation of Ios

- – source code for the runway simulation of Ios

- – draft paper

Bio: Heidi Howard is currently studying towards PhD at the Cambridge University, Computer Lab, under the supervision of Prof. Jon Crowcroft. Her research interest is fault-tolerance, consistency and consensus in modern distributed systems. In 2014, she received her BA in Computer Science from Pembroke College at the Cambridge University. Heidi has also previously worked on topics such as middlebox traversal, DNS , privacy preserving systems and wireless community networks.

This talk is part of the Computer Laboratory Systems Research Group Seminar series.

Tell a friend about this talk:

This talk is included in these lists:

Note that ex-directory lists are not shown.


© 2006-2024, University of Cambridge. Contact Us | Help and Documentation | Privacy and Publicity