On Tue, 21 Aug 2012, Dimitri Maziuk wrote: > On 8/21/2012 2:28 AM, Lars Marowsky-Bree wrote: >> On 2012-08-21T00:22:00, Dimitri Maziuk <[email protected]> wrote: >> >>> CLUSTERIP > ... >> Some customers want to do this for a simple load balancing solution that >> doesn't require additional load balancing hosts. It's not the worst idea >> for 2-3 nodes, especially with an asymmetric connection profile (i.e., >> little data in, lots of data out). > > <shrug/> I fail to see the advantage over, say, RRDNS, or > heartbeat+ldirectord, but that's just me. (@Andrew: it is a fascinating > intellectual excersize, I didn't say it was stupid in itself.)
fewer pieces than either RRDNS or ldirectord. with RRDNS, changes to the size of the cluster (taking a system out for maintinance) can't easily be done. you have to do secondary things like moving the IP addresses to other systems. With CLUSTERIP you just reconfigure the systems and the load is now distributed differently. with ldirectord you have an extra network hop, and you have all your traffic going through one system. This is a scalability bottleneck as well as bing a separate system to configure. CLUSTERIP isn't the solution to every problem, but it works really well for many problems. >> I have a preference for active/passive clusters as long as that remains >> feasible, since that makes the architecture simpler. But once one goes >> beyond that, it's nice to have options. > > Yeah, sure. Original Poster asked what's the point of an active-active > cluster without the same ip address on all nodes and my answer is still > "tcp/ip doesn't work that way". Pointing him to clusterip, anycast, and > whatever other clever tricks people came up with is IMO the opposite of > helpful. I strongly favor sticking with active/passive clusters as well, but when you get to the point where you need more than a single box working on the problem, you need to consider the various options, and CLUSTERIP is a very strong contender. It's especially good as a simple migration from active/passive. You just add a third box to the cluster and change the configuration so that instead of your VIP moving from one box to another, it's instead shared between the systems. This is a much smaller step than setting up an external load balancer system. David Lang _______________________________________________ Linux-HA mailing list [email protected] http://lists.linux-ha.org/mailman/listinfo/linux-ha See also: http://linux-ha.org/ReportingProblems
