Setting Bind so clients don't cache resolved Domain's IP's ??

Danny Mayer mayer at gis.net
Sun Dec 2 03:05:51 UTC 2001


Everyone else is racking their brains to reduce the need to make queries
of the authorative nameservers and reduce overall network loads.  You
want to INCREASE this?  Not only that, HTTP 1.1 allows the browser to
keep open the connection and fetch additional files (of any type including
jpegs) so that it does not need to reestablish the connection for each
piece of the Web page.  The setup/teardown costs of each connection
is a substantial cost of a page with lots of images and other items embedded
in it.

It you really need to do load-balancing of your Web server you should be
considering other strategies, like putting images on different servers.

What makes you think that what you want to do is a good idea?

         Danny

At 07:00 AM 12/1/01, Scott M wrote:
>Hi,
>
>Say I have a html document that has 10 images, all like <img
>src=http://www.domain.com/imageX.jpg>
>
>At the moment when I request www.domain.com, only one DNS request is made
>and the client's browser uses the returned IP to get all the data, but I
>would like it so the client does not cache the returned IP and therefore
>when ever it comes across www.domain.com in the html it has to re-query the
>DNS server. So if there is ten images and 4 links to html pages, then I want
>10 DNS queries to be made and if they choose a html page then I want it to
>look up the DNS again.
>
>Does any one know if this can be done, with TTL's or a no cache option in
>bind or something or is this something that the clients browser does?
>
>Thanks,
>Scott



More information about the bind-users mailing list