Bert Hubert finaly blogs (again)
Ok, so Bert Hubert wrote another article and i thought: “damn, now i need to think of something to post to my blog in order not to lag behind”.
Well guess what i found something to blog about. That very same post that you can also digg btw. ;-)
What he says is in essence that for large programming projects you go through a refinement process that you have probably never planned for but nonetheless will experience just as he has with PowerDNS and that you need a userbase that does error reports and sticks with the project in order to pull it of.
Wait, users? Oh yeah thats me too… Hmm so this post is about me, err no us. Hello community.
Btw. what he says about the PowerDNS recursor is very true, it is a remarkably mature product now.
Starting with version 3.1.4 i did not experience a single software induced crash of our 20 running recursor processes where with BIND9 it was a bi-weekly annoyance. Also the switch to PowerDNS recursor cut SERVFAIL responses to half of what they were before.
Update: I need to correct myself there, median SERVFAIL responses are roughly 1/10th of what i experienced with BIND9 longterm. That may of course denote that authoritative Server operators are more on the lookout for misconfigurations now than they were ~3 years ago, but i seriously doubt that.
Second Amendment: Ok the title was corrected, Its not Bert Huber, sorry ;)
