Moderating BIND


To alleviate high costs, BIND offers the following option in the options section of named.conf:

 options {     ...     dialup yes;     heartbeat-interval 360;     ... }; 

When this option executes, BIND concentrates maintenance activities (which normally are spread out over time) to a short time interval, so it occurs all at once. This activity occurs once every heartbeat interval minute. The heartbeat interval number should be many times higher than your dial-up connection timeout. If your timeout is 15 minutes and you set your heartbeat interval to 4 hours, BIND causes as many as 6 connections in 24 hours in other words, roughly 90 minutes of connection time per day for nothing. Shortening the timeout and increasing the heartbeat interval will help decrease this.

When set up this way, BIND still originates traffic when the following occurs:

  • A user, or software, causes a lookup to be performed. This usually occurs when the user, or software, wants to speak to a machine on the outside, so the line would get connected anyway.

  • When the server is master for a zone, NOTIFY requests are sent to the slaves immediately. The disconnect timeout should be long enough that the slave has time to get back to the master and transfer the zone. The necessary timeout is several minutes for example, 3 5. In most cases, this also takes care of all necessary contact between master and slave servers with regard to refresh and retry activities. But not in all cases. Therefore, the server should be kept online for longer than the longest retry interval of your zones so no chance exists for your slave servers to expire the zone(s).



The Concise Guide to DNS and BIND
The Concise Guide to DNS and BIND
ISBN: 0789722739
EAN: 2147483647
Year: 1999
Pages: 183

flylib.com © 2008-2017.
If you may any questions please contact us: flylib@qtcs.net