My comments on the IP over X.25 MIB

["Dean D. Throop" <[e-mail]>]

After rolling in the changes from the last working group I noticed 
some things in the IP over X.25 MIB that I would like some more 
discussion about.  

Should we record the time of the last translation failure?  This 
would allow a manager to see if the last failure happened recently 
or a long time ago.  It might also allow correlation of the failure 
with other information.  To do so we would have to add objects 
to record the last EnAddrToX121LkupFlrTime and 
x121toEnaddrLkupFlrTime.  

Should we record Q-bit failures on a Per PLE (interface) basis and 
record the remote address rather than keep a per peer counter?  
Right now we record address translation failures on a per interface 
basis; shouldn't Q-bit failures be treated about the same?  Doing 
this would also allow us to capture Q-bit failures from hosts 
without entries in the peer table (for DDN hosts without any 
entries this might be real nice).  It might also require less space 
in agent memory to just keep one counter and an X121address then to 
keep a counter for every peer.  

Should we record the time of the last Q bit failure?  

Should minimum open, idle, and hold down timers be kept on a per 
PLE (interface) or per peer basis?  Per Peer allows control of 
different values for different remote peers but requires more 
memory to hold all the different values.

Should we record failures on a per system basis rather than a per 
interface table?  Rather than having everthing per interface we 
could make it global to the system; this might reduce the number of 
objects required.  

Think about these issues and let me know what your feelings.

Dean Throop		[e-mail]