[nsp] Counting bps...

[Daniel Roesen]

On Tue, Mar 18, 2003 at 01:32:46AM +0200, Volodymyr Yakovenko wrote:
> One practical suggestion - _always_ do tests of new IOSes for possible SNMP 
> counter's bugs. 

It's extremely hard to test for those, if they surface only after
weeks of production running on arbitrary interfaces on arbitrary
routers of a significant deployement base...

So you tested the image in lab, then for weeks "in the field" on
selected routers, found nothing wrong. Then you do mass deployment.
And some weeks later you're faced with significant amouts of money
lost because counters of several customers stopped working.

Really nice.


Regards,
Daniel