hi list, i have a web application running in a very reliable configuration, but hardware is old (or even inappropiate) and i have the opportunity of a hardware update so now i'm starting to see new solutions of high availability.
i dont know if i should keep the current config (which i am very confident) or switch to a newer approach (virtualization...). my web app is in php, with postgres database, for a mid range organization. it has 500 active users, but concurrent users never reach 200. i don't have hardware requeriments, just the hard is getting old. my current config is: - two pcs in failover configuration with heartbeat, and ldirectord doing load balancing. the master node has two virtual IP, one is the address of my app in the LAN, and the other is the default gw of my webservers. this config switches automatically to the slave node if master goes down. -two hp proliant dl140 g2 webservers, with the web app, which has session data in database and csync sinchronizing files uploaded by the users in regular intervals. -one pc and one proliant dl580 g3 with postgres databases, in a warm standby configuration. the switching from warm standby to active is done manually. this scenario is doing it really well, we have no downtime for problems related to the servers (except when the other dl380 died some time ago), even when we update the OS (we use debian, but probably switching to ubuntu in the next upgrade), since we do it going offline one by one. so my questions are: - do you find any weakness in my current config? - switching my config to a virtualized enviroment will give me some major advantage or improvement? any links you could provide me with information about high availability concepts and theory (probability, spof, etc), virtualization options, and maybe a little feedback about which benefits has virtualization over "traditional" configurations are very welcome. thanks in advance! -- Roberto Scattini _______________________________________________ Linux-HA mailing list Linux-HA@lists.linux-ha.org http://lists.linux-ha.org/mailman/listinfo/linux-ha See also: http://linux-ha.org/ReportingProblems