Re: FW: 64 bit counters in MPLS MIBs

[Juergen Schoenwaelder <schoenw@ibr.cs.tu-bs.de>]

>>>>> Wijnen, Bert (Bert) writes:

Bert> Well, if you want to do a GETBULK on the table, I think you will
Bert> have to deal with the gaps for the noSuch cases. And that (I
Bert> think) is not fun.

Reasonable management applications have to deal with such "gaps"
anyway so there is no added complexity here. In fact, in some
application environments, it is invisible whether a specific counter
does not exist and there might be various reasons:

(a) the agent does not implement the counter
(b) the access control makes the counter invisible
(c) the protocol version makes the counter invisible

In general, I do not expect application developers to dive into
this. I personally prefer to grab whatever counters are there and to
use the one with the highest resolution in case there are 64 and 32
bit counters counting the same thing.

What is really missing are guidelines for application developers on
how to write applications for some well-know MIBs (and probably also
some free reference implementations). Just publishing the MIBs and
letting the application developers figure out themself what they
should do with the variables puts a great burdon on them and is IMHO
part of the problem. (But I am sure I said this before in other
contexts.)

/js

-- 
Juergen Schoenwaelder    <http://www.informatik.uni-osnabrueck.de/schoenw/>