Back to search

Cornell University

Toward Robust High Performance Distributed Services

Abstract

dc:description.abstract

This thesis presents steps towards simplifying the implementation of robust high performance distributed services. First, we investigate consensus algorithms in the context of fault tolerant systems. Consensus algorithms, often a critical part of fault tolerant systems, are notoriously difficult to implement. We present a skeleton consensus algorithm that can be instantiated into several well-known consensus protocols, providing insight into the structure of consensus algorithms as well as the differences and performance tradeoffs between different algorithms. We investigate onestep Byzantine agreement algorithms which exploit contention-free situations to provide low latency performance. We present definitions of one-step algorithms and prove a lower bound on the number of processors required for such algorithms, thereby showing that our algorithm is optimal. We then generalize our investigation to k-set agreement. Second, we present RPC chains, a communication primitive that improves the performance of geodistributed enterprise applications. Distributed enterprise applications are often built as a composition of more basic services. Currently, such compositions are built using remote procedure calls (RPCs). This results in a rigid and inefficient communication pattern. RPC chains is a new ion that applies two well-known ideas, function shipping and continuations, to allow an improvement in the latency performance and reduction in the overall bandwidth usage of such geodistributed systems.

Degree

thesis:*
Name thesis:degree_name
Ph. D., Computer Science
Level thesis:degree_level
Doctor of Philosophy
Discipline thesis:degree_discipline
Computer Science
Grantor
Cornell University
Year dc:date.issued
2011

Author and committee

dc:creator, dc:contributor.*
Author dc:creator
  • Song, Yeejiun
Chair dc:contributor.chair
  • Van Renesse, Robbert
Committee members dc:contributor.committeemember
  • Huttenlocher, Daniel Peter
  • Halpern, Joseph Yehuda

Subjects

dc:subject × 3

Rights

Language dc:language.iso
en_US

Identifiers

dc:identifier.*
Dc Identifier Bibid
7955569

Chain of custody

source
Harvested from
Cornell University
Base URL
ecommons.cornell.edu/server/oai/request
Last updated
2026-11-19
Source record
OAI-PMH GetRecord
citation

Song, Yeejiun. Toward Robust High Performance Distributed Services. Doctor of Philosophy thesis, Cornell University, 2011. https://hdl.handle.net/1813/30741