To:
"Cricket Liu" <cricket@acmebw.com>, "Bruce Campbell" <bruce.campbell@apnic.net>, <dnsop@cafax.se>
Cc:
"Levon Esibov" <levone@Exchange.Microsoft.com>
From:
"Stuart Kwan" <skwan@Exchange.Microsoft.com>
Date:
Mon, 24 Apr 2000 08:25:55 -0700
content-class:
urn:content-classes:message
Sender:
owner-dnsop@cafax.se
Thread-Index:
Ab+s4WoeKoFKQpYCQCWI35X3YCKlKwBHvktA
Thread-Topic:
root server load and dynamic updates.
Subject:
RE: root server load and dynamic updates.
Title: RE: root server load and dynamic updates.
In Windows 2000, if the dynamic update client discovers that the enclosing zone is the root zone, it will (by default) NOT send the dynamic update request.
However, if the name of the enclosing zone is at least one label long (ie. "com", "edu", "arpa", etc) the client will send the update request. The request will be sent to the primary name server that is named in the SOA RR of the zone.
Levon and I would like to help investigate the possibility of Windows 2000's contributing to a load increase. Who should we contact?
Cheers,
- Stuart
-----Original Message-----
From: Cricket Liu [mailto:cricket@acmebw.com]
Sent: Saturday, April 22, 2000 10:03 PM
To: Bruce Campbell; dnsop@cafax.se
Subject: Re: root server load and dynamic updates.
> Dynamic Updates on a 'orrible scale. It appears to be that if a windows
> 2000 machine cannot find its forward and reverse name to resolve
> correctly, it will send a dynamic update for the name to the nearest found
> parent.
>
> Logically, this should *not* be the root nameservers. Well, logically,
> yes. If the logical (listed) parents are not found, then I think it walks
> back down the chain, and (possibly) to the roots.
Yup. This is basically standard behavior for dynamic
update routines, whether Microsoft's or the ISC's: find
the enclosing zone, locate its name servers and send
the dynamic update.
> As our[1] nameservers are authoritative for a few reverses[2], we tend to
> see an awful lot of these dynamic updates, which show up as 'Unauthorised
> update' etc etc.
>
> Its the equivilant of a new-born baby waking up during its first night,
> and immediately bawling down the phone to the deed poll office to change
> their name. Annoying, but it does save the parents the tramua of naming
> the babe.
I've also heard rumors that these particular dynamic
updates are coming in over TCP, which would help
explain why they're so hard on name servers. Can
anyone verify that?
cricket