WACCO and LOKO: Strong Consistency at Global Scale Public Deposited

Downloadable Content

Download PDF
Last Modified
  • March 19, 2019
  • Bethea, Darrell
    • Affiliation: College of Arts and Sciences, Department of Computer Science
  • Motivated by a vision for future global-scale services supporting frequent updates and widespread concurrent reads, we propose a scalable object-sharing system called WACCO offering strong consistency semantics. WACCO propagates read responses on a tree-based topology to satisfy broad demand and migrates objects dynamically to place them close to that demand. To demonstrate WACCO, we use it to develop a service called LOKO that could roughly encompass the current duties of the DNS and simultaneously support granular status updates (e.g., currently preferred routes) in a future Internet. We evaluate LOKO, including the performance impact of updates, migration, and fault tolerance, using both traces of DNS queries served by Akamai and traces of NFS traffic on the UNC campus. WACCO uses a novel consistency model that is both stronger than sequential consistency and more scalable than linearizability. Our results show that this model performs better in the DNS case than the NFS case because the former represents a global, shared-object system which better fits the design goals of WACCO. We evaluate two different migration techniques, one of which considers not just client-visible latency but also the budget for the network (e.g., for public and hybrid clouds) among other factors.
Date of publication
Resource type
Rights statement
  • In Copyright
  • Maggs, Bruce
  • Aikat, Jay
  • Smith, F. Donelson
  • Mao, Z. Morley
  • Reiter, Michael
  • Doctor of Philosophy
Degree granting institution
  • University of North Carolina at Chapel Hill Graduate School
Graduation year
  • 2015
Place of publication
  • Chapel Hill, NC
  • There are no restrictions to this item.

This work has no parents.