Chathuri Lanchana Rubasinghe Gunawardhana
Unobtrusive Deferred Update Stabilization for Efficient Geo-Replication
Tese submetida para provas de mestrado em Engenharia Informática e de Computadores
Instituto Superior Técnico, Universidade de Lisboa.
Geo-replication is a requirement of most cloud applications. A
fundamental problem that geo-replicated systems need to address
is how to ensure that remote updates are applied and made visible to
clients in a consistent order. For that purpose, both clients and
servers are required to maintain some form of
metadata. Unfortunately, there is a tradeoff between the amount of
metadata a system needs to maintain and the amount of concurrency
offered to local clients. Given the high costs of
managing large amounts of metadata, many practical systems opt to
serialise some updates, using some form of sequencer, which, as we will
show, may significantly reduce the throughput of the system.
In this paper we advocate an alternative approach that consists in
allowing full concurrency when processing local updates and using a
deferred local serialisation procedure, before shipping updates to remote
datacenters. This strategy allows to implement inexpensive mechanisms to
ensure system consistency requirements while avoiding intrusive effects on
update operations, a major performance limitation. We
have implemented and extensively evaluated our
approach. Experimental data shows that we outperform sequencer-based
approaches by almost an order of magnitude in the maximum achievable
throughput. Furthermore, unlike proposed sequencer-free solutions,
our approach reaches nearly optimal remote update
visibility latencies without limiting throughput.
- Unobtrusive Deferred Update Stabilization for Efficient Geo-Replication
- Chathuri Lanchana Rubasinghe Gunawardhana
- MSc Thesis. Instituto Superior Técnico,
Universidade de Lisboa.
- July, 2016.
- Available BibTeX, MSC Thesis, and extended abstract.
- Unobtrusive Deferred Update Stabilization for
- C. Gunawardhana,
M. Bravo, and L. Rodrigues.
- In Proceedings of the 2017 USENIX
Annual Technical Conference, Santa Clara (CA), USA, July 2017.