[c-nsp] IOS reliability

[Peter Rathlev]

Thank you all for the input. Especially Gary who wrote off list:

> IOS reliability, as for most software, depends to a large extent on
> how often one exercises those parts of the software that simply are
> less used and tested.  I have never run into problems with any IOS for
> the simple case of forwarding packets.
> Add in a routing protocol, or complex prioritization and queueing, or
> packet modification(s), (or Appletalk or IPX!), and one will be using
> code paths that may be less common, and more likely to have bugs or
> side effects.

I think is a very good argument and will include it in our descriptions.

> I have never seen a list of the IOS features that indicate how many
> customers are using them, and therefore how extensively tested those
> code paths are "in the real world", and the matrix of those features
> (customer uses CoPP + MQC + ISIS, etc), and what the MTBF numbers are.
> I doubt anyone has that info.

That's too bad. I guess some data mining in all the TAC cases could give
some idea about the maturity of features / feature combinations.

> In other words, my guess is is that you are on your own.  (Are you
> happier now?)

Very much so. :-) Thanks!

Regards,
Peter