[c-nsp] TACACS "emergency" password management

[Phil Mayers]

On 01/11/10 17:46, David Rothera wrote:
>
> We use it simply because if one person leaves the organization it is as
> simple as removing one user and then they no longer have access.

Sure. TACACS has a lot of plusses (pardeon the pun) we just feel 
relatively few of them are a big win for us e.g. we have a small team 
with low rate of turnover so a leaving, which is very rare, just means a 
password change, which is good practice to do often anyway.

I realise we're an outlier in this.

>
> Sure we use failover local accounts but these can only be used if the
> TACACS server is down (all three of them) and even then the local
> password is some obscure string that is stored in our CI database (one
> of the few advantages of working in an ITIL house :P)

...which is what I'm asking: how do you ensure you have fast, reliable 
access to that database during a (sufficiently large, probably rare) 
outage? How do you know you won't be blocking on availability of that 
database?

I can think of a few obvious ways; I'm just wondering what people 
actually *do* :o)