Raghtur wrote:I'm not saying that there's too many people for the server to physically handle. I'm saying that the update to the OS MIGHT have caused a programed value change limiting the number of connections or people online that the server will process at the same time. To put it simply think of the server as a call center. Now this mythical call center has 100 phone lines with 100 people to answer them, or in otherwords no waiting provided you have gotten through. Now if the server OS update has limited the number of connections the server will process it's equivalent to the mythical call center firing say 50 of the people that answer the phone lines. There's still 100 phone lines to call in to, but only 50 people to answer them, so 50 of those calls coming in need to wait.
I only came up with this idea because I and a few others have noticed from the restarts after the update that at some point between 100-200 people online seems to be the turning point for when we all start experiencing significant lag. Don't believe me, take a look back through this topic as others have posted noticing this as well after the restarts.
To summarise what he said without all the bloat: OS change could've reset a value within the OS's settings that limits the maximum amount of connections that can go through the server at once. If the value is smaller than the amount of players trying to connect (or rather, send packet requests), then some of the players or other programmed entities will have to 'wait for their turn', causing lag spikes when several hundred players are online at once and can't all be served at once.
Though, limiting the amount of incoming connections a server can take is common practice, especially in defense against DDoS attacks. I'm still convinced that the problem lies mostly in the server-side coding, and not so much in the actual server specs. Though of course, a SSD with a single backup HDD for redundancy would speed up the I/O, as others have said.