> >But if they're designed so that they have problems with allocating 64 KB >of memory, there's something about the design that's wrong IMO.
It's not really a design issue, it is a cost issue... The implementations used in these boxes are quite configurable -- the person who integrates the stack into the box can make all sorts of choices about how many buffers are available of what sizes, etc. This restriction would require that the integrator allocate at least one 64K buffer (or enough buffer space in a chained buffer system) to re-assemble a 64K packet. These systems are very cost-sensitive, and vendors are very careful about limiting the memory size, especially the R/W memory size, of these devices. They will typically complain about anything that "wastes" more than ~16K of memory. What are the major down sides of limiting the MTU on these tunnels? Margaret -------------------------------------------------------------------- IETF IPng Working Group Mailing List IPng Home Page: http://playground.sun.com/ipng FTP archive: ftp://playground.sun.com/pub/ipng Direct all administrative requests to [EMAIL PROTECTED] --------------------------------------------------------------------
