All things related to multicast

[Hugh LaMaster <>]

On Sun, 18 Jul 1999, Mark Fullmer wrote:

> > I see nothing malevolent about NTP multicast traffic,
> > and, in fact, as long as most of the people advertising 
> 
> What's wrong with peering with a stratum 1 server and 
> distributing the multicast or broadcast locally?  Listening

Nothing.  Unfortunately, for reasons I don't understand,
most campuses and networks don't have stratum-1 servers.
I assume that as people become more sensitive to performance
measurement and (taken broadcast) "QoS" issues which benefit 
from accurate network-wide time, more organizations will
start supplying such clocks.  (Not to mention that some
security measures require accurate clocks, and that 
distributed filesystem users (e.g. NFS) find it convenient
if their workstations and servers are closely synchronized.)

> to random NTP mcast sources around the globe seems problematic
> at best.

If there are millions of stratum-16 sources, the benefit
is lacking, I agree.  But, with a modest number of high-quality
stratum-1 servers (unfortunately, most of the multicast servers 
seem to be stratum-2 right now, true, but, since we are talking
"in principle"), I see no harm in listening to a few high-quality
"random" sources.  NTP is very good at selecting the best source
among a number of good ones.  Basically, with NTP, up to
reasonable numbers, "the more the merrier".  Also, even for
organizations which have such a stratum-1 server, that server 
sometimes it breaks and nobody notices for a while, and then,
the "NTP" guy is on vacation or has just gone to work for 
somebody else or whatever.  Having backup sources available turns
out to be a good thing in such circumstances.

> We peer our routers with local clocks and configure all the
> LAN interfaces with
> 
>  ntp broadcast destination 255.255.255.255

Every campus/enterprise LAN should do that.  Unfortunately,
most don't.


Basically, I see no harm in allowing multicast NTP traffic
from a reasonable number of high-quality servers.  Since
these are the only (e.g. MSDP) sources, they don't cause
a router state explosion and there really isn't that much
traffic.


$0.02 worth of opinions from:
Hugh LaMaster




--
 Hugh LaMaster, M/S 233-21,    Email: 

 NASA Ames Research Center     Or:    

 Moffett Field, CA 94035-1000  Or:    

 Phone: 650/604-1056           Disc:  Unofficial, personal *opinion*.