Re: Choosing length of lease

["Michael J. Lewis" <[e-mail]>]

A very unscientific study of traffic at my installation showed that DHCP
took .05% maximum of LAN traffic.  (This occurred Monday morning during
the after weekend reboot cycle.)  Two DHCP servers were in use at the 
time one with a seven day lease and the other a 30 day lease.  Normal
DHCP traffic ranged from .01% to 0%.  (FYI, I do mean .01% not one 
per cent.)  


Mark Sirota wrote:
> 
> John M. Wobus wrote:
> > (1) At your own site, how long do you make leases and why did you
> >     choose this time?
> 
> Initially, our leases in the residential dorms were made one year long.
> That way, students were guaranteed to have the same address all year,
> which is very important to them.  Furthermore,  those few who stay
> in the dorms year-round get the same address indefinitely.
> 
> We have since reduced it to four months, long enough to cover the
> summer.  If a student renews near the end of the academic year, and
> gets back early, and lives in the same dorm, they'll get the same
> address the following year, but we will survive the big September
> crunch.
> 
> In non-Resnet networks, we generally do one-day leases, with a maximum
> of one week.
> 
> > (2) Are there lease-lengths that you would consider to be a
> >     lower-bound or upper-bound for any ALL practical uses?  If so,
> >     what would you be trying to avoid by staying above/below this
> >     bound.
> 
> One concern I have about the very short lease times that some people
> have reported is, what about load on the server and the network
> to which the server is connected?  In a large installation, with
> thousands of addresses being served by a single central server, this
> seems like it could be a consideration.
> 
> On the flip side, it looks like just about everything communicates with
> the server when it boots, even if the lease is fresh, so maybe short
> leases (as long as they are one day or more) don't introduce that much
> extra traffic.
> --
> Mark Sirota, Network Systems Engineer
> University of Pennsylvania, Information Systems and Computing
> [e-mail], 215/573-7214