(no subject)

John Hascall john at iastate.edu
Sat Feb 9 19:08:50 UTC 2008


We are starting to see issues with the time to process
a DHCP request which seem to be tied to the time to
complete dynamic DNS updates.

I ran a tcpdump of dynamic dns update traffic from our
DHCP server for 200 minutes on Friday (11:08 - 14:28)
then I computed how long each request took and then
grouped, averaged and sorted the times by zone.

The results show that our largest and most updated
zone takes by far the longest for an update to
complete.  Taking on average over 100 times longer
than all other zones to update.  Some stats:

Average                                       Number   Rank
Seconds/                                        of      by
Update                  Zone                 Updates   Time
--------     ----------------------------    -------   ----
0.537275             student.iastate.edu.      12796      1
0.005323     <average of all other zones>      14387
0.037834         22.186.129.in-addr.arpa.         21      2
0.017874         48.186.129.in-addr.arpa.         32      3
0.001409                engl.iastate.edu.         35    151
0.001384                 las.iastate.edu.         29    152
0.001380                  hs.iastate.edu.         69    153
0.001364               psych.iastate.edu.         15    154
0.001039             physics.iastate.edu.         12    155

What I am wondering is: Is there some issue of scale
with respect to zone size or number of updates that
might be causing this and what can we do about it?

Also, any news on when DDNS updates might be done
asynchronously by DHCP?

PS, We are running both DHCP and DNS on recent machines
(Dual 3.00GHz Xeon, 2GB, hardware mirrored SATA drives, GB NICs)
with recent versions of ICS's DHCP and DNS software
(3.1.0 and 9.3.4 respectively).

More information about the dhcp-users mailing list