Ping Spikes
Hi Im a hardcore gamer. . play alot of FPS-games on-line My ISP deliver 10Mbit over a LAN connection and my ping looks real good at first glance, I have like 1-3ms inside their network and i get like 15ms to server inside my country.
Hi
Im a "hardcore" gamer ..play alot of FPS-games on-line
My ISP deliver 10Mbit over a LAN connection and my ping looks real good at first glance, I have like 1-3ms inside their network and i get like 15ms to server inside my country.
But yet the game feels like runnin in mud so i started to investigate my connection.
At first they told me i had alot of CRC-errors and collisions -Wow I thought, thats it... so they fixed it (turned out that they put me on half-duplex on my port) but the game on-line still sucked.
I start using PingPlotter and pretty fast found out that my 3 first jump is spiking from 1ms up to 220ms every now and then.
Its like those 3 jump takes turn on spiking, if the first and third is ok then the second spikes and vice versa so its like a forest of spikes that ruin on-line gaming.
I might add that i have 10Mbit DEDICATED so im not sharing it.
The fourth jump and those after are perfect, they spike like top 4ms over a 24hour monitoring.
They have worked on this problem for quite a long time but to no avail.
I wonder if there is anyone on this forum whit insight to what could be the cause of this.
On my end I have switched between 3 diffrent NIC´s, 5 TP cables and tried all Duplex settings etc.
Is it the third router jump thats causing the problem to reflect back to the second and first jumps?
Or is it the first thats causing problems for the second and third, and if so -Why does it stop after the third jump?
They say that they have no limitation set for UDP or any ports that are used for gaming.
I hope there is some hardcore technician here who could help me, help them
Im a "hardcore" gamer ..play alot of FPS-games on-line
My ISP deliver 10Mbit over a LAN connection and my ping looks real good at first glance, I have like 1-3ms inside their network and i get like 15ms to server inside my country.
But yet the game feels like runnin in mud so i started to investigate my connection.
At first they told me i had alot of CRC-errors and collisions -Wow I thought, thats it... so they fixed it (turned out that they put me on half-duplex on my port) but the game on-line still sucked.
I start using PingPlotter and pretty fast found out that my 3 first jump is spiking from 1ms up to 220ms every now and then.
Its like those 3 jump takes turn on spiking, if the first and third is ok then the second spikes and vice versa so its like a forest of spikes that ruin on-line gaming.
I might add that i have 10Mbit DEDICATED so im not sharing it.
The fourth jump and those after are perfect, they spike like top 4ms over a 24hour monitoring.
They have worked on this problem for quite a long time but to no avail.
I wonder if there is anyone on this forum whit insight to what could be the cause of this.
On my end I have switched between 3 diffrent NIC´s, 5 TP cables and tried all Duplex settings etc.
Is it the third router jump thats causing the problem to reflect back to the second and first jumps?
Or is it the first thats causing problems for the second and third, and if so -Why does it stop after the third jump?
They say that they have no limitation set for UDP or any ports that are used for gaming.
I hope there is some hardcore technician here who could help me, help them
Participate on our website and join the conversation
This topic is archived. New comments cannot be posted and votes cannot be cast.
Responses to this topic
Hi
Thanks for the reply
I gotta read and digest what you wrote here, I´m in a hurry now so i repsond quickly and add a pic/log from PingPlotter from me that you could study.
The 3 graphical segment at the bottom is the 3 first jump and the spike they make.
The fourth segment is..you guessed right, the fourth jump..and its totally flat, no jumps at all http://hem.bredband.net/b211419/30-september2.gif
You can see the MAX spikes on the 3 first jumps and you could see on the graphical segment that it is the third jump that makes the most "noise".
By the way, heres the link to PingPlotter http://www.pingplotter.com/ .
Do you know what they could do to resolve the problem?
I check in later
Thanks for the reply
I gotta read and digest what you wrote here, I´m in a hurry now so i repsond quickly and add a pic/log from PingPlotter from me that you could study.
The 3 graphical segment at the bottom is the 3 first jump and the spike they make.
The fourth segment is..you guessed right, the fourth jump..and its totally flat, no jumps at all http://hem.bredband.net/b211419/30-september2.gif
You can see the MAX spikes on the 3 first jumps and you could see on the graphical segment that it is the third jump that makes the most "noise".
By the way, heres the link to PingPlotter http://www.pingplotter.com/ .
Do you know what they could do to resolve the problem?
I check in later
Hmm, the fact that the ping delays jump could actually mean some sort of network traffic delay/packet collision on the ISP's network or even where their network connects to the backbone.
How many jumps are there between your ISP and their backbone provider(s) ?!? What this means is that they give the end user a connection and then in turn routes your packets to one or more backbone providers like UUnet and/or Verio for instance. There are a number of backbone providers out there with varying amounts of capacity as well.
It's also possible that you ISP takes many more hops and/or a funky route to get to their backbone provider(s).
Have to done a trace route to the end server(s) you frequently play on ?!? This will give you more info on what route the packets take to get to their destination from your connection.
What you appear to be experiencing is latency or lag as it's called and of course FPS games tend to be the most sensitive to this kind of thing
How many jumps are there between your ISP and their backbone provider(s) ?!? What this means is that they give the end user a connection and then in turn routes your packets to one or more backbone providers like UUnet and/or Verio for instance. There are a number of backbone providers out there with varying amounts of capacity as well.
It's also possible that you ISP takes many more hops and/or a funky route to get to their backbone provider(s).
Have to done a trace route to the end server(s) you frequently play on ?!? This will give you more info on what route the packets take to get to their destination from your connection.
What you appear to be experiencing is latency or lag as it's called and of course FPS games tend to be the most sensitive to this kind of thing
Quote:How many jumps are there between your ISP and their backbone provider(s) ?!? What this means is that they give the end user a connection and then in turn routes your packets to one or more backbone providers like UUnet and/or Verio for instance. There are a number of backbone providers out there with varying amounts of capacity as well.
Have to done a trace route to the end server(s) you frequently play on ?!? This will give you more info on what route the packets take to get to their destination from your connection.
Hi
Thanx for youre input
There is 3 jumps to the backbone as you can see on this one >> http://hem.bredband.net/b211419/30-september2.gif .
Doesnt this one qualify as a TraceRoute, or could I get more information about the connection in another way?
I always pass through the 4 first jumps wherever I go, so Im´stucked whit those spikes always.
This is by the way the route to one of my favorite game-server and you can see that if the pingspikes wasnt there I would have a total of 5ms ping to it.
Have to done a trace route to the end server(s) you frequently play on ?!? This will give you more info on what route the packets take to get to their destination from your connection.
Hi
Thanx for youre input
There is 3 jumps to the backbone as you can see on this one >> http://hem.bredband.net/b211419/30-september2.gif .
Doesnt this one qualify as a TraceRoute, or could I get more information about the connection in another way?
I always pass through the 4 first jumps wherever I go, so Im´stucked whit those spikes always.
This is by the way the route to one of my favorite game-server and you can see that if the pingspikes wasnt there I would have a total of 5ms ping to it.
Quote:If you would like to try the one I created back in 1996-1997 & have modded for the added features Lotus (member here) requested, the IP-UserName-Password combination for d/l via FTP is:
Server = emu-pdx.myftp.org
Port = 2077
User = leechmenow
Pass = ntcompatible
* Needs testers, should be 100% ready-to-rock-n-roll & all that, but tests never hurt... if you don't & are happy w/ that shareware, then no biggie!
(Mine's a freebie offering though & has alot of features now that the other one may not include...)
I will try it as soon i get a FTP client installed
Quote:Again, it's probably what I theorize above about it being nested between 2 other ICMP responses & this forces the IP stack to budge in the process queue since it is a driver & also up its CPU priority &/or usage.
The DOS/DDOS attack method I list above may be a better way for you to understand this... read it when you can
My knowledge about the OSI-model and TCP/IP stack is kinda limited, to say the least
Quote:That might also be carrier noise (rx, tx, cx) too, another thing to consider as well... like when you download files, it tends to go up
I never downloaded anything while I made those PingPlotter logs, just to avoid misreadings. If thats what you ment.
Quote:Patch the OS most likely... & check the arp cache as per the article in the MS url link I put up above also.
There are registry hacks that can restrict the number of dropped packets & repeat replies as well, I think this is the one from here:
HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Services\Tcpip\Parameters
EnableICMPRedirect & TCP1320Opts & TcpMaxDupAcks (especially this last one)
I looked for those strings in the registry but I only found EnableICMPRedirect and it was set to 1, however I downloaded WinGuides TweakManager and whit that program I could make those changes you proposed.
But sadly it didnt make any diffrence at all.
The ARP-thing I havent looked into yet, will do.
When you say Patch the OS i guess you mean the lastest updates and servicepack and drivers, which I have the newest of all
Quote:If this is, however, restricted to a PARTICULAR SERVER, (e.g.-> not diff. ones you ping, but one you always ping & see it spike up) then the problem may be at that server/node on the WAN itself! Things like latency being high in return trip from it, or packet loss manifesting itself ONLY ON THAT NODE would be server-side & place to concentrate on
This happends on every single server I ping to, doesnt matter which part of my country or if I ping a server abroad
I appreciate all your input here
Server = emu-pdx.myftp.org
Port = 2077
User = leechmenow
Pass = ntcompatible
* Needs testers, should be 100% ready-to-rock-n-roll & all that, but tests never hurt... if you don't & are happy w/ that shareware, then no biggie!
(Mine's a freebie offering though & has alot of features now that the other one may not include...)
I will try it as soon i get a FTP client installed
Quote:Again, it's probably what I theorize above about it being nested between 2 other ICMP responses & this forces the IP stack to budge in the process queue since it is a driver & also up its CPU priority &/or usage.
The DOS/DDOS attack method I list above may be a better way for you to understand this... read it when you can
My knowledge about the OSI-model and TCP/IP stack is kinda limited, to say the least
Quote:That might also be carrier noise (rx, tx, cx) too, another thing to consider as well... like when you download files, it tends to go up
I never downloaded anything while I made those PingPlotter logs, just to avoid misreadings. If thats what you ment.
Quote:Patch the OS most likely... & check the arp cache as per the article in the MS url link I put up above also.
There are registry hacks that can restrict the number of dropped packets & repeat replies as well, I think this is the one from here:
HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Services\Tcpip\Parameters
EnableICMPRedirect & TCP1320Opts & TcpMaxDupAcks (especially this last one)
I looked for those strings in the registry but I only found EnableICMPRedirect and it was set to 1, however I downloaded WinGuides TweakManager and whit that program I could make those changes you proposed.
But sadly it didnt make any diffrence at all.
The ARP-thing I havent looked into yet, will do.
When you say Patch the OS i guess you mean the lastest updates and servicepack and drivers, which I have the newest of all
Quote:If this is, however, restricted to a PARTICULAR SERVER, (e.g.-> not diff. ones you ping, but one you always ping & see it spike up) then the problem may be at that server/node on the WAN itself! Things like latency being high in return trip from it, or packet loss manifesting itself ONLY ON THAT NODE would be server-side & place to concentrate on
This happends on every single server I ping to, doesnt matter which part of my country or if I ping a server abroad
I appreciate all your input here
maybe the ISP needs to simply restart one of it's routers - or even replace it?
