loftar wrote:Serejai wrote:A game like this causes a LOT of strain on the hard drive because of all the read/write data it requires to update tiles.
It's not as if the changing of a tile translates to a hard drive IOP, however. The tiles, as well as all other data, are kept entirely in memory and only saved back to disk when the map area is being unloaded. The same can be said for most, if not all, other actions that can be performed by players. Haven doesn't actually use that much disk I/O at all.
Serejai wrote:Personally I would go with a Xeon minimum (and if they're using an i7 that would be one of the causes of random lag spikes) but overall the I/O is the bottleneck for a server like this and a 7200RPM HDD just doesn't cut it.
It doesn't? Since the BSDDB patch and kernel upgrade, I think the server has been handling exceptionally well. There has been virtually no server-side lag even with 1400+ players continuously online, and no sign of near contention, either.
Serejai wrote:Now, I don't know which specific hosting package is being used... but the lowest one comes with 16GB of RAM which should be adequate to run this server (although 32GB would be better) and an i7.
For the record, the hosting package we ordered isn't on the site any longer. The server has 12 GB of RAM and an i7 920.
On the other hand, it should be noted that Hetzner has previously been reusing used hard drives when replacing crashed ones. I don't quite know what to think about that, because there really is no reason to just assume that they are bad (they could have come from decommissioned servers or other harmless uses), and I can certainly see that it would be a pointless expense from their side to buy entirely new disks every time they build a new server or any time someone complains of dubious troubles with their hard drives. On the other hand, it's not as if they couldn't be bad, either. It's a good question whether the failures have been caused by the hard drives being latently bad or if there's something I'm doing.
On the other hand, I did notice that they have now added an option to have a completely new drive inserted instead, but at cost. I perused that option for one of these replacements, and the other disk also happened to be completely new (even though I hadn't requested it), so the server is, at least, now finally running on new disks. Let's hope that has an effect.
Either way, I can't really see that there's anything in particular that I would be doing that should be especially strenuous on the drives. Once this is all over and dealt with, I do plan to file a complaint with Hetzner about the horrible MTBF exhibited thus far (6 drives in less than 3 years is hard to ignore), but I would like to take this opportunity to mention that, apart from the disks, I have enjoyed Hetzner quite a lot. They're very "hands-off", their network works well, their support crew is reasonable and quite accommodating, their automatic management options are very nice, and their quite cheap for all that; so if I were to change providers, I don't really know to what.