Very high WoW pings between 7pm and 9pm

[#1]

I've noticed in just the last week that my pings in WoW have skyrocketed.  Normally, they are 40-50ms.  Sunday 250ms.  Tonight... 400ms.
Also, maybe related, when doing a traceroute at that time to the WoW server there appears to be a fair amount of packet loss registered between the plusnet routers.  Also my first hop router is showing about 30ms instead of 20ms. However, looking at the packet loss graphs, I can't see any obvious problems.
I'm not getting any errors on the lower level ADSL connection... very low error counts etc. and the graph using DMT looks very clean.  My exchange (Almondsbury) doesn't report having any contention issues.
When this started to happen last Sunday or so, I disconnected my router and re-connected leaving the lowest level ADSL connected and got a different gateway and my pings dropped down back to sensible levels.  However, staying on that gateway tonight my pings were back up in the 250-400ms range again, and trying to swap gateways didn't seem to fix it.
Has WoW somehow been de-registered as a top priority game?
Thanks!
-- gyre --

[#2]

When was the last time you saw it working perfectly?  If you can pin it down to a day, that would help with seeing it if correlates with any updates to our systems.

[#3]

If it helps, I play WoW most evenings and haven't noticed any Plusnet-specific latency.
Various realms have been suffering from latency internally (the Bloodlust battlegroup was taken offline last weekend to try to sort out localised latency there) so it may not be Plusnet related.
When you next have high latency, try creating a character on a realm in a different battlegroup and see if it has any affect.
B.

[#4]

It was working superbly for the last few weeks. After Blizzard had changed their client at 2.3.3 to use the 'nagle fix', I was seeing raid pings of under 55ms, and normal pings of 38-45-ish.
I believe it was Sunday around 5 O'Clock when it started all going wrong.  One minute I was on a 50ms ping, then I rebooted the router for a different reason and on reconnection my ping started to skyrocket.
When I connected and disconnected a few times from the gateway I eventually found one that had 50-55ms pings again.  However, trying that approach last night didn't sort the problem.
One of my daughters was on EU-Aggramar at the same time as I was on EU-Bronze Dragonflight and was also complaining about the latency last night.
I've noticed that WoW's ping measurements seems to be very heavily influenced by packet loss.  Any loss at all sends the ping measurement sky-high.
Thanks
-- gyre --

[#5]

A few weeks ago I posted this fix on the ElitistJerks forums and it worked well for most people. It was supposed to be fixed in 2.3.2, however it seems that the Nagle Algorithm fix in patch 2.3.2 isn't as effective as expected for most players, and if you're in need of a lower ping this is still the best way to go.
Read this first
1 - It might have side effects, like slowing down your download speed or affecting your performance on other software. (In my experience, it doesn't change anything but you have been warned)
2 - If you're not sure about what you're doing, just don't do it.
2.1 - Seriously, if you don't know what you're doing, don't do it.
3 - If you don't have the HKEY_LOCAL_MACHINE\SOFTWARE\Microsoft\MSMQ\ directory, you can download and apply this file to your registry.
4 - Windows Vista users might want to check this post too.
The Method
1 - TcpAckFrequency - NOTE if you are running Windows Vista this setting will not have any affect. This works fine under Windows XP
Go to Start->Run and type "regedit" and click Ok, to bring up the registry editor
Then find:
HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Services\Tcpip\Parameters\Interfaces\
There will be multiple NIC interfaces listed in there, and you need to find the one you use to connect to the internet.
There will be several interfaces listed (they have long names like {7DBA6DCA-FFE8-4002-A28F-4D2B57AE8383}). Click each one in turn, and the right hand pane will show lots of settings in it. On one of the interfaces, you should see your computer IP address listed. 
When you've found the correct interface, right-click in the right hand pane and Add a new DWORD value, name it TcpAckFrequency, then right click the entry and click Modify and assign a value of 1.
You can change it back to 2 (default) at a later stage if it affects your other TCP application performance. It tells Windows how many TCP packets to wait before sending an ACK. If the value is 1, windows will send ACK every time it receives a TCP package.
2 - TCPNoDelay
This one is pretty simple (Discussed here)
Go to Start->Run and type "regedit" and click Ok, to bring up the registry editor
Then find:
HKEY_LOCAL_MACHINE\SOFTWARE\Microsoft\MSMQ\Parameters
Right-click in the right hand pane and Add a new DWORD value, name it TCPNoDelay, then right click the entry and click Modify and assign a value of 1.
Click Ok and close the registry editor, then reboot your PC.
Basically, this fix is properly deactivating the Nagle algorithm to improve your ping.
Once you're done with the technical stuff (it probably won't take more than a minute), you can finally get your reward.

In this case the fix was used to decrease a very high ping, it also seems to work very well on low (~100ms) ping for some people (lowered to ~20 or ~30ms).
This may help, but I'll just reiterate:  If you don't know what you're doing, don't do it.
B.

[#6]

Actually, I still have both of those fixes in place.
I applied them a while back after seeing them on the EJ forum (thank you!) and never bothered backing them out, since that PC is dedicated for WoW.
So yup, I've had uber-low pings for quite a few weeks now.  Great for being a main tank.  Up until Sunday that is
-- gyre --

[#7]

In that case, I'd be interested to see the results of a Pingplotter Trace over your connection during a laggy session.
Instead of tracing the login server (at eu.login.worldofwarcraft.com) can I suggest you do the following:
From a command prompt (if in Vista choose Run As Administrator):
netstat -n -p tcp -b | more
Run down the list of connected ports.  You're looking for a line similar to:
TCP    192.168.0.1:1234      80.239.181.107:3724    ESTABLISHED    4040[WoW.exe]
The important part is that it's opened by the WoW.exe file, and the destination port is 3724.
Take the IP address listed in the above example (80.239.181.107) and run a Pingplotter on it during periods of latency.
The reasoning behind this is that the login servers are all hosted in Blizzards Paris datacenter at Villacoublay.  The realm servers are hosted in several locations in France and Germany, so the route to your realm may actually be different to the route to the login server.
Once you have a pingplotter trace (of at least 5 mins or so) save the file and post it here and I'll take a look at it.
Also, it's worthwhile running a wireshark trace and submitting the file for Dave "The Guru" Tomlinson to check to see whether the TOS field is appropriately classifying the traffic as Gold.
B.

[#8]

Thanks.  I'll do both of those next time the pings go high.  Probably tonight down BT
-- gyre --

[#9]

Ahh, we're only just starting Hyjal ourselves, but good luck
B.

[#10]

Thanks.  We stopped for Christmas as 3 of our 6 tanks left.  But we've now attuned replacements.
As far as the TOS goes on the IP header:
Differentiated Services Field: 0xa0 (DSCP 0x28: Class Selector 5; ECN: 0x00)
Seems to be what wireshark is saying at this moment.  Will check when things get bad, although I'd have thought it would be unlikely to change.
-- gyre --

[#11]

Ok, it definately seems like it's being categorised as Titanium by the Ellacoyas which means it should get top priority over the Plusnet network.
Seems like the Pingplotter trace is the next step to try, but that will have to be done during periods of high latency.
(Out of interest, could you post a tracert to your realm server now - I'd be interested to see the routing)
B.

[#12]

Currently, the pings are fine.  29-35ms.  But then I expect that this time of day.
Here is the current traceroute to the world server while I'm in nagrand.  Note that the 99ms pings on hop2 is a standard glitch for speedtouch routers
Tracing route to 80-239-185-21.customer.teliacarrier.com [80.239.185.21]
over a maximum of 30 hops:
  1    <1 ms    <1 ms    <1 ms  10.0.0.6
  2    12 ms    99 ms    99 ms  speedtouch.lan [10.0.0.138]
  3    23 ms    23 ms    22 ms  lo0.plusnet.thn-ag1.plus.net [195.166.128.101]
  4    22 ms    23 ms    23 ms  vl703.thn-gw1.plus.net [84.92.4.241]
  5    23 ms    22 ms    23 ms  ldn-b4-link.telia.net [213.248.84.201]
  6    31 ms    23 ms    23 ms  ldn-bb1-link.telia.net [80.91.249.77]
  7    29 ms    31 ms    29 ms  prs-bb1-pos6-0-0.telia.net [213.248.64.9]
  8    29 ms    30 ms    30 ms  prs-b1-link.telia.net [80.91.250.253]
  9    32 ms    30 ms    30 ms  prs-rosy-ks51-link.telia.net [80.91.250.66]

[#13]

I checked out the Gateway Status graphs which give you a good indication of congested gateways and nothing seems amiss at all.
So, it will boil down to the Pingplotter trace from when you do have latency
B.

[#14]

Yup, I also checked the graphs last night for each gateway I connected to, and nothing seemed amiss.
I'll run a pingplotter tonight if/when things go bad.  Maybe that will shed some light on the problem.
Thanks
-- gyre --

[#15]

I just had another thought - it might be worthwhile checking your usage on the VMBU page and see if it has gone up considerably recently - tying in with your increased pings.
Maybe some new software on one of your computers is misbehaving - a full spyware and virus scan would be an idea.
Probably something you've already checked, but worthwhile doing if you haven't.
B.