The IP Over NBMA (ION) Archive[Date Prev][Date Next][Thread Prev][Thread Next] [Date Index][Thread Index][Author Index][Subject Index] Request for progress to Informational RFC
Hi Grenville, Joel > [..] > >>As a second point (minor), in describing the MARS scaling problems there is > >>a reference to the processing time increasing as the database size > >>increases. Given that no order of magnitude is given, the reader might > >>be lead to think that this is a linear increase. It is of course > >>distinctly sublinear, to the point that one might wonder at its relevance. > > Fair point. I have no gut feel for the relative relevance of intra-MARS > processing time as a function of group membership database size. I suppose the intra-MARS processing time would be heavily dependent on the specific implementation. More specifically, it would be dependent on how the database in the MARS is organized and managed. So what may be more interesting is to determine the control message load on the MARS as a function of the cluster size or the group size. Having established a relation here, we can then compute the load on the MARS for specific implementations. Regards, Rajesh. -- Rajesh R. Talpade (404)-315-1227 taddy@cc.gatech.edu http://www.cc.gatech.edu/computing/Telecomm/people/Phd/taddy
|
|