Connection statistics gadget.

[Glenolas]

<p><span><p>Sorry for the cross post, but I was unaware of this forum when I posted in the general area.</p><p>There's a connection statistics gadget in game, usually toggled by the F11 key.</p><p>Can anyone offer a definitive answer as to what is contained in <span style="color: #00ffff;"><strong>the small aqua colored number</strong></span> on the top row to the immediate right of the large "frame rate" number?</p><p>TYIA</p></span></p>

[kdmorse]

<p>Light Blue (cyan) is in game ping time in milliseconds.  The time it takes a packet to get from within the bowels of the client, to the bowels of the server, and back.  Related to, but not the same as your network ping time, which is just host to host.  Higher numbers contribute to the lag of every action you take.  Lower is better.</p><p>Yellow is bytes per second between you and the server</p><p>Green is frames per second on your client.</p><p>Red is packet loss.  If it's anything other than a flat 0, you've got problems.</p><p>-Ken</p>

[Brook]

<p>Ping time from client to server and back.</p><p>He beat me to it.</p>

[Glenolas]

<p><cite>Ynnek@Kithicor wrote:</cite></p><blockquote><p>Light Blue (cyan) is in game ping time in milliseconds.  The time it takes a packet to get from within the bowels of the client, to the bowels of the server, and back.  Related to, but not the same as your network ping time, which is just host to host.  Higher numbers contribute to the lag of every action you take.  Lower is better.</p></blockquote><p>I used the question to start a conversation on the matter.  I am exploring server lag, and believe the aqua number to be a measure of it.   </p><p>I understand computer to IP (and return) ping, and router  issues along the way, getting on and off the internet backbone, etc.   Let's call it "true" ping.  I can measure that ping from the command line.</p><p>I live in Florida, so San Deigo round trip at the speed of light (theoretical best speed) would be about 27msec, and my computer to San Diego zoo ping hangs around 50 msec, which is excellent.   I use SD zoo because it's a steady measure of the distance portion of the aqua number, if the aqua number actually contains true ping in it's makeup.   By chosing a neutral site I eliminate my router, my IP router, the internet backbone routers, and getting off the backbone and into San Deigo from contentious discussion.   They are all included n the 50 msec standard ping measurement.</p><p>My aqua number hangs between 170msec and 230msec, and is more commonly on the upper end.   It is far greater than my true ping, and it's  fluctuation does not seem to be related to my true ping (which hangs within 3 msec of base while the aqua number varies over it's 60msec  minute to minute range. </p><p>Further, it does not seem to be packet related, (which I'd not expect it to), but that rules out client lag of any serious amount.  I.e. zero packets to/from = negligible client activity other than running the gadget itself.    </p><p>Making matters even more confusing is that if I talk telephonically to people in San Deigo who are also in game, their numbers vary over the same range mine are, at the same time.    That sort of rules out even my distance related true ping being included in the aqua number.   The further away from San Deigo I am, the worse the number should be, but it doesn't seem to follow that notion.   My numbers should be on the order of 50 msec bigger than theirs, and they are not.</p><p>So now you're ready for the next level question.   If the aqua number doesn't include what we all know and love as computer to IP ping,  or if that ping is a minor portion of the aqua number, where does the rest of the number come from?  Is it a measure of  front door of SOE to server, rattle around inside, then back out to the front door of SOE?   (long winded definition of server lag).   If not, what is it?</p>

[Glenolas]

<p><cite>Brook wrote:</cite></p><blockquote><p>Ping time from client to server and back.</p><p>He beat me to it.</p></blockquote><p>I asked the question because while your answer is the commonly accepted one, I don't believe it.   Please see my further discussion above.</p><p>I'm willing to have it explained to me, but needs further depth than client to server and back, in the sense the regular measure of ping, because it far exceeds that ping. </p>

[kdmorse]

<p>Short answer: I believe there is truth to what you suspect, but how much truth we'll probably never know.</p><p>Long answer:</p><p>First comment, why ping the Zoo?  Ping your server.. This gives you the network, IP layer, round trip time between your system, and the server you are talking to.  A much more reasonable starting number.  It might be very different, or of course it could be the same.  But at least it's more accurate.  We can take that as the network round trip time.</p><p>Then, there's the time getting in and out of your client.  This is a nontrivial amount of work, especialy since the packets are actually objects, and get serialized/deserialized on every trip.  There's very real CPU time involved locally, and it will be at least partly dependant on your system's speed.</p><p>And of course, there's getting in and out of the server.  The same serialization/deserialization applies.  Now, assume you're the only one logged on, you could get a nice baseline as to how much time it takes to get in and out of an empty, idle server (it won't be 0ms).</p><p>Add all the above up, and you have a baseline best-case in-game ping (which we will probably never know).  On top of that, a congested server can slow things down, increasing the round trip time - exactly the effect you're hinting at.  But how much severe congestion contributes to how much increased response time is going to be a vauge correlation at best. Especially since we don't know how eq2 server event queue is managed.</p><p>Additionally, due to quirks in windows archetecture, ms level timing in a 32-bit application is actualy harder than you might think.  Unless the developer went out of his way to use a high performance timer, the accuracy (depending on what timer they used) could easilly be +/- 52ms, or +/- 15.9ms. </p><p>So, it all comes back to my short answer.  I believe there is truth to what you're implying.  But how much, or how truely usefull it is, we may never know.   (Someone who plays on lots of different servers I suppose may actually be able to draw better conclusions than I do, on my one backwater server).  But since you mentioned it, I'll start paying attention to the blue numbers a bit more just for giggles.</p><p>-Ken</p>

[SkunkCabbage]

<p>I'm trying to figure out what appears to be a major degradation in game performance since switching ISPs.  I am getting terrible lags and lag spikes; periodically my display simply freezes for what feels like an eternity.  My ping number seems to hover around 250 or so; but the bottom number in the display varies from 0 to over 5,000.  I have made no changes to my computer, but have not had performance problems like this since the Sentinel's Fate expansion "settled down".  I'm afraid that I'm not a hardware person; does anyone have any ideas?  I'm thinking of returning to my old ISP but would really like to have a less foggy idea of what's happening.</p><p>Thank you!</p>