A few things...
1.) AI is required to be turned on or the DB doesnt like it and stops sorting properly.
2,) Monsters too.
2.) Shipyard errors (bad shiplist) cause the same DB instability and add size to the DB.
3.) Forfeits can add to disrupted DB as the DB doesnt understand how to handle them.
4.) Player disconnects and bad mission reporting, same as above.
5.) More AI litters the map from forfeits and player disconnects and improper mission reporting leading to AI spam in the DB requiring a DB clean to fix.
There are a few others, I forget.
Now, once the DB starts to swell, a few other factors come into play.
1.) The processor will start to spike.
2.) The larger the player load, the more processor usage.
3.) The larger the shiplist, the more processor usage.
4.) The faster the shipyards run, the more processor usage.
5.) Forfeits can add to disrupted DB as the DB doesnt understand how to handle them.
6.) Player disconnects and bad mission reporting, same as above.
There are a few others, I forget.
The tolerance of how much the processor can handle depends on many factors, but the bottom line is once the DB swells, the other factors contribute to an inevitable processor spike and once it reaches 100%, the processer cannot keep the DB ordered and it swells exponentially soon causing a burp or crash. Once restarted, the processor is under low load and is able to start up again. But it repeats the steps above and the same burp happens again once the processor reaches 100%.
Your processor handles all these problems quite well because it is so fast. But even it too reaches a point where the only thing that will allow it to run for a while w/o a burp is a DB clean. But if a DB can be kept small from avoiding the things that make it swell, then depending on the processor speed of the server, the next step is to see just how high the player load can go!
50, 60, 70?
I'd put my money on 70, not that now we'd ever get that many cuz of the bumps that drove off peeps from the past till this point here, but if you remember articfires, he had 50-60 peeps on his. No DB clean back then. But if you remember, he used stock missions, mostly stock gf settings and stock shiplist. Unfortunately, there were other factors that caught up with the DB in the end and corrupted it permanently. In its last days, player count started to take its toll and it held less and less and started to burp more...
I'd say we are 90% there to higher player numbers and running 12+hrs. straight w/o a crash. Bonk and I have done extensive testing this server and our tests affirm our conclusions for the most part.
Bonk, did I miss anything?
Topic: ***DB Stability Achieved! SGO4 has been up 24hrs. SOLID! (Read 2128 times)