> >I know what I can do as root and I take the responsibility seriously.
> You're one person, how many work there?  Are they all as good you?  And
how
> are we supposed to know that?

  You know, I've been listening intently to the conversionations regarding
this, and Paul, you just made the first comment that really hit home..

  Many of the conversations had here is, 'Why shouldn't I have root access?
I can handle it, I'm not gonna do anything stupid'.  That's not the point.
Only people who *absolutely need* root access should have it.  Again, to
clarify, I'm not talking a development lab here, I'm talking live production
systems.  Yes, this includes desktops..  root means god.  They can do
practically anything.  The problem is, alot of network security depends on
the security of the local box.  Once I have root, I can do alot of things
that ordinarily, due to security limitations I can't.  Such as interface low
level IP packet info, allowing me to mess with the overall protocol being
used on a network.  I can sniff network traffic.  Heck, I can start
accessing raw TTY consoles and installing a sniffer program, pulling yet
more ssh entered root passwords for other machines..

  Now, I *AM* a software engineer.  I can see where they're coming from.
Being an engineer, I truely, truely, TRUELY *HATE* the idea of it.  It's is
a *VERY* large pain in the arse to have an admin have to give me every itty
bitty detail of permision to do each and every thing I want to do EVERY time
I need to do it on our production boxes.  But I can understand why.

  All it takes for a root password to be compromised is for someone to
happen to see you type the damned password.  That's what it means when you
say 'Trust no one".  Your not even saying that that INDIVIDUAL may do
something, it's that there is no way to 100% gaurantee that the knowledge he
has may not be, even accidently, compromised.  Every 'instance' of knowledge
of that data provides for a certain margin of possible compromise of the
resource.  By only allowing a select few, possibly only 2 or 3, access, you
greatly reduce the possibility of a compromise.  Let's say the margin is
.01% per individual.  If 3 individuals have access, it's cumulative of .03%.
Now, if an R&D department of 30 has access, your likelyhood is now .3%, *TEN
TIMES GREATER*.

  Now, I'm not speaking for development labs, becouse by nature labs have
holes that have been opened up to allow for development, testing, debugging,
or just plain dicking around with the newest kernel or latest toys.  It's a
cadaver lab..  ;-P  And yes, in this environent, people use it as
'production', aka, I develope on the same network that I read email, etc.
But this is the R&D production environment, so if we do majorly screwup, as
we eventually (and have) done, sure, we could flood the network something
nasty..  Feature of being in an R&D environent is that we can indeed see how
something interacts in a quasi-real environment.

  I guess I can just see it from both sides, and when it comes down to it,
it's a pain in the ass to protect my companies assets, but we gotta do what
we gotta do.  That's one of the jobs on a Systems Administrator.  Sure, it's
nice to say that SysAdmins sole job is to make the Engineers job living
hell, but they're doing their jobs..



**********************************************************
To unsubscribe from this list, send mail to
[EMAIL PROTECTED] with the following text in the
*body* (*not* the subject line) of the letter:
unsubscribe gnhlug
**********************************************************

Reply via email to