dslreports logo
 
    All Forums Hot Topics Gallery
spc
uniqs
11

dslx_nick
ISP Employee
join:2011-12-24
Chatsworth, CA

dslx_nick to Rockafella21

ISP Employee

to Rockafella21

Re: Problem with inconsistant download speed.

Sounds like we already have a ticket open to resolve this for you, but feel free to PM me your account info or ticket number when you get a chance.

Rockafella21
join:2010-04-08
New Castle, PA

Rockafella21

Member

Click for full size
Congestion/Latency
Click for full size
pingplotter_http
I have setup pingplotter with WinPcap to use TCP packets on port 80 to prove there is in fact a congestion issue going on with my connection. While it is true that some routers do prioritize ICMP traffic which makes tracerts less reliable using TCP packets eliminates this situation. As you can see from the image below there is ALOT of congestion taking place on hops 4,5, and 6. Now that we have the diagnosis lets work on the cure. Also lets get it worked on today.

****Edit I have included a screen shot to show that the packets used are definatly TCP HTTP packets so this rules out the ICMP prioritizing.

dslx_nick
ISP Employee
join:2011-12-24
Chatsworth, CA

dslx_nick

ISP Employee

Er... it's entirely possible I'm missing something very obvious, but I'm just not seeing any congestion going on there in that first screencap. All of the hops have green indications on the left, since they are 0-200 ms... 13-35ms is not what I'd consider congested.

Rockafella21
join:2010-04-08
New Castle, PA

1 edit

Rockafella21

Member

The 13-35ms that you are referring to is showing you the latency at that given time in the screenshot. If you look at the arrows it shows you how high the latency has gotten at its max. You can also see how many times the latency has increased by looking at the graphs to the right of the arrow. See how the first 2 hops without the arrows are very stable with only one spike? Also this test was done before 5pm and the latency was already getting high.
Rockafella21

Rockafella21 to dslx_nick

Member

to dslx_nick
I have just ran a line quality test that proves the same thing that pingplotter was showing me. As I told you I am noticing a problem at the last Verizon hop right before being passed off to Alter.net. The IP address of the hop that is giving me problems is 130.81.28.204. In pingplotter I am showing when this hop is acting up it starts affecting all the hops after it. This results in major packet loss at the last hop. Here is the line quality test results »/pingt ··· /2886398.
Rockafella21

1 edit

Rockafella21 to dslx_nick

Member

to dslx_nick
So I take it ping of 330ms to a local server and speeds of 3.8mbps is normal for the 7.1mb/768kb speed tier? Latency of 900ms to a server that is less than 100 miles away. Sure that indicates a problem I would think.
»www.speedtest.net/result ··· 1857.png
»www.speedtest.net/result ··· 5524.png
»www.speedtest.net/result ··· 7411.png
»www.speedtest.net/result ··· 7425.png
»www.speedtest.net/result ··· 9492.png
»www.speedtest.net/result ··· 0747.png
I guess playing the Xbox 360 online is not allowed on a DSL Extreme connection. They will say everything is fine on their end. Can someone please get this problem resolved sometime soon? This is getting ridiculous!
Rockafella21

Rockafella21 to dslx_nick

Member

to dslx_nick
Click for full size
Here you go. Now you can see exactly what I'm talking about. All the way down to the last hop is affected.
Rockafella21

1 edit

Rockafella21 to dslx_nick

Member

to dslx_nick
Also right now the default gateway is producing 50% packet loss. The packet loss is happening off and on. I am also showing latency spikes at the default gateway as well. I am testing to two different sites at the moment dslreports and speedguide.net. It is 6:15Pm right now. I am going to continue to test until 9Pm or so. According to the information I am seeing a circuit change is needed. This is the same thing that has happened in the past. What happens is I get moved to a location with few customers. Verizon sees how well the connection is running so they move hundreds of other people over who are having problems. It never fails. They create their own problems.

dslx_nick
ISP Employee
join:2011-12-24
Chatsworth, CA

dslx_nick to Rockafella21

ISP Employee

to Rockafella21
I'm looking at the chart, and it looks like the massive spikes in congestion are starting between hop 1 (192.168.x.x) and hop 2 (10.x.x.x). That would suggest the source of latency may be your own (in-house) router. Try removing that device, then running another pingplotter test and let's see what the results from that are.

Rockafella21
join:2010-04-08
New Castle, PA

Rockafella21

Member

Its not the router because when I run the line quality tests it shows 0 packet loss on that hop. Also if it was my router it would be producing the problem constantly and not just doing peak hours. We both know that.

dslx_nick
ISP Employee
join:2011-12-24
Chatsworth, CA

dslx_nick

ISP Employee

10.x.x.x is an internal IP address. Do you have some other device, such as a VOIP or external firewall?

We really need to test without whatever that 10.x.x.x device is, because the max readings are spiking to 300+ starting from it. The average column readings are only about 27, round trip.