I have just setup 2 new servers per ISPConfig's Cluster tutorial on HowToForge.com, with Debian, and everything appears to be working fine. How does one test the load balancing, though, to make sure it's working properly? I have read a lot online about load balancing and there are talks of different methods, DNS RR seeming to be the least best way, apache with the mod_proxy seeming one of the better routes without spending any money, or setting up an actual 3rd server to act as a load balancer. The ISPConfig tutorial mentions none of these options, so does it just work? Or does it use the apache mod_proxy method automatically? I just am trying to figure out a way to make sure this is working properly. Also, I have these 2 servers, per the tutorial, in master/master mysql replication. Now I am looking to use Amazon's DynamoDB so this shouldn't be an issue, but how would one after the fact, add server 3, 4, 5,...? Master/Master doesn't sound like the best for additional servers, but in this situation with just 2, I guess it's fine. I am wanting to make sure we don't run in to problems down the road when we need to add additional servers to our cluster to expand for our project.
The tutorial keeps two servers in sync (mail, web, database). You can setup a load-balancer in front of the two servers or just use DNS-RR. If you add additional web-servers they can use your current 2 db-servers or run mysql local or - if you have 2 additional web-servers - they can also replicate theire databases. You can also setup a mysql-cluster
Thank you for your reply, that helps a lot. I had a feeling there was something else I needed to do. The tutorial makes no mention of "what next" once you've setup the cluster. My 2 servers are clustering apache, DNS, mysql, email, just as the tutorial describes. I am not using DNS or email, so really just apache and mysql. Although I may not even end up using MySQL and go for Amazon's DynamoDB. It unfortunately means rewriting my program though to use DynamoDB instead of MySQL. So pondering that decision. If you had to ballpark, how many simultaneous users do you think the 2 server cluster can handle? Right now I have 4GB RAM and 2 virtual processors which are at about 2GHz each if I understand the QEMU CPU descriptions correctly. No other websites are being hosted but my application. I can easily upgrade the virtual instances to have 8GB RAM and 4 virtual CPUs at a moments notice, so very flexible to provide more resources.
