Originally Posted by
MiaShino
This is incredibly common for the type of deployment they are hosting.
Example;
For emulated servers I ran we used three separate physical systems for most sub 1000 user worlds.
Server Bank A (Login and queuing) - Bank of two physical servers that handled account data/logging in/queuing into world shards. These route you to world server of choice. Single dual core was sufficient.
Server Bank B (item database) - Learned most quickly that item lookup consumed too much IO and needed own separate server. One box with ten drives in raid scalable should situation require. Multi processor setup here per box, accessing that many drives is noooo joke. Required many pci-e lanes.
Server Bank C (world server) - Handled player position and mob position updates. Was also running many Lua scripts for mob - boss AI. Eight gigabyte setup, needed much much more of the memory then other servers.
Server Bank D (core database) - Multi server SQL database for parsing all non account player data such as character/race/level/name/etc. This is where we would have executed the lottery upon as this had many more cpu cycles to spare then other banks.
No conspiracy here only standard operating procedure.