tag:blogger.com,1999:blog-15622618.post112458013556375392..comments2007-04-14T05:50:56.164-07:00Devananda vdvhttp://www.blogger.com/profile/03378341125039609409noreply@blogger.comBlogger2125tag:blogger.com,1999:blog-15622618.post-1134582501628932512005-12-14T09:48:00.000-08:002005-12-14T09:48:00.000-08:00Not remembering the exact tool that we used back then, I've been trying to find names of software that accomplish RRDNS with automatic failover, and have been coming up short. One could use <A HREF="http://powerdns.com/" REL="nofollow">Power DNS</A> with an additional script, monitoring SNMP, that updates the underlying DNS-database when it notices trouble, but I don't think that's what we used.<BR/><BR/>Since you've called my attention to this, I've been thinking - what would really best suit this internal layer, so there can be failover among the SQL nodes in the cluster? It's not going to be RRDNS, since even in best-case scenarios, there is inherently some delay (at most the full TTL in the DNS record). The best solution is going to be a proper load balancer, something akin to the <A HREF="http://www.linuxvirtualserver.org/" REL="nofollow"> Linux Virtual Server</A> project (assuming you want to stay away from commercial products).Devananda vdvhttp://www.blogger.com/profile/17104394848350685084noreply@blogger.comtag:blogger.com,1999:blog-15622618.post-1134353917625748372005-12-11T18:18:00.000-08:002005-12-11T18:18:00.000-08:00in your diagram you say <B>round robin DNS automaticallly updates to redirect traffic if a server fails</B>. What do you use to detect server failure and remove DNS's A record from the list?Henryhttp://universalplatform.comnoreply@blogger.com