Next: Summary
Up: Reliable Multicast Transport
Previous: Congestion Avoidance for Reliable
Handley has proposed a framework within which we can consider
reliable multicast protocol work. In this, he describes the
categories of parameters of the system:
-
Number of sources: 1-to-n, n-to-m, or other.
-
Transmitter start time: is there one?; is there one for each receiver
marked at join time?
-
Real time-Ness - what are the latency constraints on delivery of data
from source to one, any or all receivers?
-
consistency
He lists as non factors both congestion control and
ordering, since
both can be done orthogonally to the above categorisation.
He then groups applications into five main areas:
- 1.
- Bulk Transfer (VR database, Usenet news etc)
-
1-many
-
relaxed real time
-
synchronised start time
-
file (object) level consistency
- 2.
- Live data feeds (stock exchange etc)
-
1 source
-
no start time
-
quasi real time
-
may need synchronised consistency
- 3.
- Resilient (streamed server data - like RTSP from video/audio from web site)
-
1 data source
-
no start time
-
stream consumed in near real time
-
non consistency
- 4.
- shared applications (wb, nte etc)
-
distributed data set
-
all sites fold most
-
relaxed consistency
-
many sources of changed
- 5.
- Hybrid - some applications (in general, the Distributed Interactive
Simulation systems) may be a hybrid of all of the above.
This is discussed further in chapter eight where we look at some
applications that require particular reliable multicast semantics.
Next: Summary
Up: Reliable Multicast Transport
Previous: Congestion Avoidance for Reliable
Jon CROWCROFT
1998-12-03