[Beowulf] Fwd: NIS limitations question

Mark Hahn hahn at physics.mcmaster.ca
Thu Feb 9 09:17:43 PST 2006


> > > I belive i have seen on this maling list*, and other internet fourms** some
> > > limitation of NIS, but i have failed to find a documented limiation from
> > > SUN, or from the various linux distrubutions, did any one try to research
> > > the scalability of NIS servers?
> > 
> > The standard answer, if you only rarely push, is to make every client
> > a slave.
> 
> The less violent solution is to simply run nscd (name service caching
> daemon) on all clients - that will take a lot of the load off of both
> your NIS and DNS servers   :)

I have deep admiration for DNS, and quite a lot of scorn for 
various other systems that try to do similar things, poorly.
for instance, LDAP works, but that's the best you can say for it.
imagine if the LDAP folk had thought of how to use DNS as a 
directory infrastructure (but alas, they were x500 recidivists ;)

observe that DNS provides a general mechanism for providing 
a synchronized database, including security, that could easily be 
used to structure a user directory, including all the usual passwd 
fields, ssh pubkeys, arbitrary site-specific stuff.  DNS has 
well-established caching, lease-like TTLs, round-robin behavior, 
delegation, secure updates, etc.  and you already have it in place.

the best argument against this, I suppose, is that existing DNS 
implementations (bind/named) are not exactly structured to make this 
easy, and are certainly not intended for this kind of use.  then again,
it's astonishingly easy to write a DNS server from scratch...

regards, mark hahn.




More information about the Beowulf mailing list