Joined: Feb 16, 2003
From: portland, oregon, usa
|Posted: 08-06-2018 17:12  |
I took some measurements. If you want to avoid spikes in lag (10% slower 50-60% of the time, 4x slower around 2% of the time), use a wired connection to play mud.
I'd felt like I'd experienced some bad lag in recent weeks, so I sat down and wrote some code to make some measurements. I'm on the west coast of the US, so it's unlikely that you'll have the same measurements.
I measured lag by logging in, and then sending a single keypress at the Option prompt and timing how long until mud echoed it back. So it's an approximation of the time between entering a command in mud and the game telling you what happened. I took 250 measurements after a brief warm up period, and then recorded the 2nd, 25th, 50th, 75th, and 98th percentile times (respectively: fastest 2%, fastest 25%, median, slowest 25%, slowest 2%).
Here's what I found: (I added _ to better align the columns)
Wifi__ mudii.co.uk 157 159 160 162 660 (2nd-25th-50th-75th-98th percentiles)
Wired mudii.co.uk 156 156 171 172 188 (2nd-25th-50th-75th-98th percentiles)
Wifi__ mud2.com_ _90 _96 185 194 702 (2nd-25th-50th-75th-98th percentiles)
Wired mud2.com_ _78 _94 _94 234 250 (2nd-25th-50th-75th-98th percentiles)
So what does this tell us?
* mud2.com can be almost 2x faster for me. However, mud2.com oscillates between periods of 80ms response times and 230ms response times. The cyclic patches of 3x slow downs will feel pretty jarring.
* wifi achieves the same speed as wired around 40-50% of the time. The rest of the time is fractionally slower. 2% of the time wifi is significantly slower than wired (around 4x slower).
* mudii.co.uk is far more consistent with it's lag. The only significant variations were introduced by wifi. My best guess is that this is because mudii.co.uk server is running on a bare metal server, and that mud2.com is running in either a virtual machine or is shared with things other than mud2.
What's suspect about this analysis?
* I took measurements once, and the speeds on the internet is pretty variable over time.
* My measurements were taken in my spot in the US. It's hard to generalize that to anywhere else in the world.
* Wifi performance is a complicated mess unto itself. All sorts of factors can affect wifi lag time. I have tuned my wifi setup quite a bit, and I have a moderately fancy wifi router and wifi adapter. So if anything my guess is that the common wifi scenario will be worse than mine.
* I don't know the specifics of the internet connectivity I have. I'm using a DOCSIS 3 cable-modem. Folks with a fiber connection to the internet may have much better lag times.
* I'm using a direct 10Gb ethernet connection to my router. It's possible that older technology (e.g. ethernet over the electric mains) may perform worse.
Any questions? Any thoughts? I'm happy to re-run or adapt my measurements if folks have any input. I've also been entertaining the idea of building the lag time monitoring into mudiiclient. I could add it as another line chart in the log files, and as an element in the status bar. I'd also be interested to know what kind of lag other folks experience. Some grouchy part of my brain imagines that folks in London are getting 60ms lag times or better to mudii.co.uk.
[ This Message was edited by: Havoc on 08-06-2018 17:29 ]