Ryan Malayter wrote: > On 3/30/07, Rob Janssen <[EMAIL PROTECTED]> wrote: >> They are just a user. The fact that they cause a problem is because of >> a limitation in our system, not because of their use. >> We need to move to better DNS servers that do not need a static zonefile >> that is reloaded every so often. Then the Turks will just get >> distributed over all servers and nobody will notice anymore. > > Is the pool using BIND, or djbdns? > > It would seem to me that a database-driven system such as MyDNS would > suit the pool far better. Also, a database-driven system would make > future enhancements like geo-location or bandwidth-based weighting far > easier. > > Although given the spotty security record of MyDNS so far, I am not > sure it is the best choice. What other open database-driven DNS > systems are out there? > > The load balancing feature of MyDNS would be an improvement over what we have now. (i.e. it could dynamically serve a large collection of available servers with different priority classes)
I don't know what the load on the DNS servers is, and if performance/efficiency is an important concern when designing things. I think what you really need is not a database-driven DNS but a "code-driven" DNS, where you can write your own backend functions (e.g. in Perl or Python) that are called for every query and can cook a response depending on many factors (including geo-location). Of course a database can be used to conveniently store server data, but a table of active servers might just as well be kept in RAM. PowerDNS could be a possibility. Rob _______________________________________________ timekeepers mailing list [email protected] https://fortytwo.ch/mailman/cgi-bin/listinfo/timekeepers
