BYOND Version:479
Operating System:Windows Server 2003 R2
Web Browser:Firefox 3.6.13
Applies to:Dream Daemon
Status: Unverified

Thus far we've been unable to verify or reproduce this bug. Or, it has been observed but it cannot be triggered with a reliable test case. You can help us out by editing your report or adding a comment with more information.
BYOND is only using about 10% of the available network speeds on my server. I have done some testing, and the server does seem quite capable of its supposed speeds on both upload and download. PlanetDownloadSpeed.png PlanetUploadSpeed.png
BYOND itself, on the other hand, seems to inconsistently limit itself. Even when I was downloading resources, the speed never exceeded what is shown in the screen-shots. And takes me about 5x longer than it should. Meaning I'm only downloading at about 200kbps, instead of my possible 1,000kpbs.
However, in previous months, I have seen that monitor program report BYOND as high as 2,500+, so the limit is rather odd. I have also seen various BYOND resources download at ~1,000kpbs, so I can't say BYOND is at fault. It may also be possible that those other resources were hosted on a web-server with high speeds, and not downloaded directly from a BYOND server.
I have multiple DD instances hosting multiple servers of various games, several of the same game, about 14 servers in total. The DD numbers shown in those screen-shots are the combined total for all servers.
I'm wondering how all this works on BYOND's end, if its linked to the website in any way? or if anyone knows of a limit within windows itself, which may be slowing certain programs?
My best guess is that it has to do with how BYOND "chunks" packets, but we'd have to investigate to see for sure.

Can you construct a simple test that downloads different sized files and track the data? That would help us get some insight here.
Downloads them to a player using ftp()?
I'm not really sure what you're comparing there for input and output. DD's only work in the way of inbound bandwidth is when the pager sends it a message, or when it's receiving info from players; the latter will be highly inconsistent with each other in terms of speed, so you may see a slow average speed there. The outgoing speed will ultimately be bound by your network and is going to be lower than the inbound speed in most cases, but output to each player will again depend somewhat on their connection.

That's not to say that BYOND's packet chunking isn't an issue or can't be improved on, though. I'm just really not sure what those graphs represent--in particular it would help to know what sorts of activity are represented in each.
The server has a 100MB connection both up and down.
I'm not sure what graphs you mean, if you're talking about the screen-shots of the NetLimiter program, then I'm not sure whats confusing. The left column is inbound (downloading) bandwidth speed - this is notably low on the DD instances. The right column is for outbound (uploading), which seems to be forcing some kind of inconsistent limit onto itself.
The screen-shots are mainly there to show that internet explorer is capable of utilizing (almost) the entirety of the net speed available on the server, but that BYOND is not.
Also, I made the point that if only 10MB are being used on average, then when I log in to the server, it should have plenty of speed available for me to download at my full 10MB, but was only getting about 2, with no noticeable increase in the reported usage. Which means everything else was slowed down to accommodate my resource download?
I was just thinking a test case where you have 1) a big resource file (like an image used for your title screen) and 2) a bunch of little resource files, and compare how long they take to transmit vs how long they should take. This is something we could test but since you've got some monitoring in place, it would help out a lot. I'm not sure we can really improve on things but this is the first place we'd check.
Moving this to Unverified for the time being until we have more info. More detailed information about the bandwidth consumption or a test case would be helpful.