This paper explores integrating self-tuning updates and
sequential consistency to enable transparent replication
of large-scale information dissemination services. We
focus our attention on information dissemination services,
a class of service where updates occur at an origin
server and reads occur at a number of replicas, and
our data replication system supports transparent replication
by providing two crucial properties: (1) sequential
consistency to avoid introducing anomalous behavior
to increasingly complex services and (2) self-tuning
transmission of updates to maximize performance and
availability given available system resources. We meet
these aggressive consistency and self-tuning goals using
a novel architecture that (1) pushes invalidations on reliable
FIFO network channels, (2) pushes updates on unreliable,
priority-ordered, low-priority network channels,
and (3) carefully schedules the application of invalidations
and updates at each replica. Our analysis of simulations
and our evaluation of a prototype implementation
support the hypothesis that it is feasible to provide transparent
replication for information dissemination applications.
For example, in simulations our system’s performance
is a factor of three to four faster than a demandbased
system for a wide range of configurations.