I am a seventh year CS Ph.D. student at Princeton University in the S* Network Systems group advised by Prof. Wyatt Lloyd and Prof. Ethan Katz-Bassett (Columbia University). Before moving to Princeton I was a member of the University of Southern California's Networked Systems Lab. I am interested in distributed systems and networking in general.
I received a B.A. in Mathematical Sciences from Rutgers University - Camden in 2016.
My current research, with collaborator Theano Stavrinos, focuses on centralized sequencers in distributed systems. A centralized sequencer is a server that hands out sequence numbers to clients in a distributed system, which use the sequence numbers to order their operations. Multi-sequencing, a recent development, extends sequencers to allow consistent cross-shard transactions. Multi-sequencing uses sequence spaces, logically independent sequences of strictly increasing integers, where, typically, each sequence space is mapped to a shard. Previous works' multi-sequencers provide a noncontiguous multi-sequence abstraction which can contain holes in the sequence — integers in the sequence that are not used — introducing complexity for the service developer. Our work focuses on providing a stronger abstraction, a contiguous multi-sequence, which assigns exactly one operation to every integer in each increasing sequence space such that each sequence space has no holes. The contiguous multi-sequence abstraction is easy for distributed service developers to reason about while being flexible enough for developers to specialize execution of their operations for better performance. We implement the abstraction in a multi-sequencing subsystem that is also the first scalable multi-sequencer.
Haonan Lu, Christopher Hodsdon, Khiem Ngo, Shuai Mu, Wyatt Lloyd. "The SNOW Theorem and Latency-Optimal Read-Only Transactions"
OSDI 2016, Savannah, GA, November 2-4, 2016.
[paper]
[slides]
[talk (Haonan Lu)]
e-mail: chodsdon at princeton edu