dslreports logo
 
    All Forums Hot Topics Gallery
spc
Search similar:


uniqs
1098
Phil_C
join:2013-06-30
Garner, NC

1 edit

Phil_C

Member

[Southeast] DSL performance degradation

We began having problems with our DSL Lite service back around the 25th of Jamuary. The service began feeling much more sluggish than usual, and I noticed a behavior during file downloads that I did not recall seeing before. Very often during a file download the data transfer completely stops for varying periods of time, often for five to ten seconds, and then continues on. While downloading a 3 MB file I will usually see an interruption in the data transfer occur at least once, and sometimes as many as four times.

Here are the results of some tests that I ran recently:

Test #1:

$ date && web100clt -n ndt.itcc.unc.edu -l
Sat Jun 29 10:30:57 EDT 2013
Testing network path for configuration and performance problems -- Using IPv4 address
Checking for Middleboxes . . . . . . . . . . . . . . . . . . Done
checking for firewalls . . . . . . . . . . . . . . . . . . . Done
running 10s outbound test (client to server) . . . . . 139.00 kb/s
running 10s inbound test (server to client) . . . . . . 531.08 kb/s
sending meta information to server . . . . . Done
Your host is connected to a Cable/DSL modem
Information: Other network traffic is congesting the link
Information [C2S]: Packet queuing detected: 22.22% (local buffers)
Server 'ndt.itcc.unc.edu' is not behind a firewall. [Connection to the ephemeral port was successful]
Client is probably behind a firewall. [Connection to the ephemeral port failed]

------ Web100 Detailed Analysis ------

Web100 reports the Round trip time = 136.14 msec;the Packet size = 1440 Bytes; and
There were 28 packets retransmitted, 98 duplicate acks received, and 104 SACK blocks received
Packets arrived out-of-order 29.79% of the time.
The connection stalled 2 times due to packet loss.
The connection was idle 0.56 seconds (5.60%) of the time.
This connection is receiver limited 25.15% of the time.
This connection is network limited 74.27% of the time.
Excessive packet loss is impacting your performance, check the auto-negotiate function on your local PC and network switch

Web100 reports TCP negotiated the optional Performance Settings to:
RFC 2018 Selective Acknowledgment: ON
RFC 896 Nagle Algorithm: ON
RFC 3168 Explicit Congestion Notification: OFF
RFC 1323 Time Stamping: ON
RFC 1323 Window Scaling: ON; Scaling Factors - Server=4, Client=10
The theoretical network limit is 0.56 Mbps
The NDT server has a 109 KByte buffer which limits the throughput to 6.28 Mbps
Your PC/Workstation has a 101 KByte buffer which limits the throughput to 5.78 Mbps
The network based flow control limits the throughput to 2.42 Mbps

Client Data reports link is ' 2', Client Acks report link is ' 2'
Server Data reports link is ' 2', Server Acks report link is ' 2'
Information: Network Middlebox is modifying MSS variable (changed to 1452)
Server IP addresses are preserved End-to-End
Information: Network Address Translation (NAT) box is modifying the Client's IP address
Server says [98.69.xxx.xx] but Client says [ 192.168.1.5]

-----

Test #2:

$ date && web100clt -n ndt.itcc.unc.edu -l
Sat Jun 29 17:31:42 EDT 2013
Testing network path for configuration and performance problems -- Using IPv4 address
Checking for Middleboxes . . . . . . . . . . . . . . . . . . Done
checking for firewalls . . . . . . . . . . . . . . . . . . . Done
running 10s outbound test (client to server) . . . . . 139.00 kb/s
running 10s inbound test (server to client) . . . . . . 35.29 kb/s
sending meta information to server . . . . . Done
Your host is connected to a Cable/DSL modem
Information [C2S]: Packet queuing detected: 31.42% (local buffers)
Information [S2C]: Packet queuing detected: 35.83% (local buffers)
Server 'ndt.itcc.unc.edu' is not behind a firewall. [Connection to the ephemeral port was successful]
Client is probably behind a firewall. [Connection to the ephemeral port failed]

------ Web100 Detailed Analysis ------

Web100 reports the Round trip time = 154.00 msec;the Packet size = 1440 Bytes; and
There were 22 packets retransmitted, 8 duplicate acks received, and 8 SACK blocks received
Packets arrived out-of-order 17.39% of the time.
The connection stalled 1 times due to packet loss.
The connection was idle 0.56 seconds (2.67%) of the time.
This connection is receiver limited 95.22% of the time.
This connection is network limited 4.50% of the time.
Excessive packet loss is impacting your performance, check the auto-negotiate function on your local PC and network switch

Web100 reports TCP negotiated the optional Performance Settings to:
RFC 2018 Selective Acknowledgment: ON
RFC 896 Nagle Algorithm: ON
RFC 3168 Explicit Congestion Notification: OFF
RFC 1323 Time Stamping: ON
RFC 1323 Window Scaling: OFF
The theoretical network limit is 0.29 Mbps
The NDT server has a 109 KByte buffer which limits the throughput to 5.55 Mbps
Your PC/Workstation has a 41 KByte buffer which limits the throughput to 2.09 Mbps
The network based flow control limits the throughput to 2.14 Mbps

Client Data reports link is ' 2', Client Acks report link is ' 2'
Server Data reports link is ' 2', Server Acks report link is ' 2'
Information: Network Middlebox is modifying MSS variable (changed to 1452)
Server IP addresses are preserved End-to-End
Information: Network Address Translation (NAT) box is modifying the Client's IP address
Server says [98.69.xxx.xx] but Client says [ 192.168.1.5]

----------

So far I've run several dozen of the above tests using servers located all over the country (and several other countries as well). Upload speeds have been very consistent, while download speeds have been all over the place.

The modem/router currently in use is a new D-Link DSL-2540B. When the problem originally started we were using the original modem -- a Westell model E90-610030-06 -- in bridged mode with a D-Link DI-804HV router attached to it. Switching to the new modem/router did not fix the problem.

Some stats from the status page of the D-Link DSL-2540B:

Downstream Line Rate (Kbps): 3552
Upstream Line Rate (Kbps): 384
 

Here are some more stats that I obtained by using the original Westell modem:

DSL Statistics
 
Line State                Up
Modulation                G.DMT Mode
Data Path                 Interleaved
 
Transceiver Information
                              Down Stream Path Up Stream Path
DSL Sync Rate (Kbits/Sec)     3552             384
SN Margin (db)                16.0             12.0
Line Attenuation (db)         46.0             28.0
CRC Errors                    0                2
 

Phil

Hayward0
K A R - 1 2 0 C
Premium Member
join:2000-07-13
Key West, FL

Hayward0

Premium Member

If you are talking about the basic 798k service 590k is a bit low but not horribly off if not constant. Remember there are over head/error check bits that are like 8% of speed.
Phil_C
join:2013-06-30
Garner, NC

Phil_C

Member

Hi Hayward,
said by Hayward0:

If you are talking about the basic 798k service 590k is a bit low but not horribly off if not constant. Remember there are over head/error check bits that are like 8% of speed.

Focusing on that one number won't give you an accurate picture of what is going on because you're looking at a snapshot in time that is only ten seconds long.

Here are all of the test results (summary containing speed tests):

running 10s outbound test (client to server) . . . . . 141.00 kb/s
running 10s inbound test (server to client) . . . . . . 49.87 kb/s

running 10s outbound test (client to server) . . . . . 144.00 kb/s
running 10s inbound test (server to client) . . . . . . 647.19 kb/s

running 10s outbound test (client to server) . . . . . 137.00 kb/s
running 10s inbound test (server to client) . . . . . . 107.16 kb/s

running 10s outbound test (client to server) . . . . . 145.00 kb/s
running 10s inbound test (server to client) . . . . . . 545.08 kb/s

running 10s outbound test (client to server) . . . . . 139.00 kb/s
running 10s inbound test (server to client) . . . . . . 287.97 kb/s

running 10s outbound test (client to server) . . . . . 146.00 kb/s
running 10s inbound test (server to client) . . . . . . 354.20 kb/s

running 10s outbound test (client to server) . . . . . 144.00 kb/s
running 10s inbound test (server to client) . . . . . . 535.12 kb/s

running 10s outbound test (client to server) . . . . . 146.00 kb/s
running 10s inbound test (server to client) . . . . . . 249.53 kb/s

running 10s outbound test (client to server) . . . . . 147.00 kb/s
running 10s inbound test (server to client) . . . . . . 87.73 kb/s

running 10s outbound test (client to server) . . . . . 145.00 kb/s
running 10s inbound test (server to client) . . . . . . 559.44 kb/s

running 10s outbound test (client to server) . . . . . 139.00 kb/s
running 10s inbound test (server to client) . . . . . . 531.08 kb/s

running 10s outbound test (client to server) . . . . . 139.00 kb/s
running 10s inbound test (server to client) . . . . . . 216.62 kb/s

running 10s outbound test (client to server) . . . . . 139.00 kb/s
running 10s inbound test (server to client) . . . . . . 94.88 kb/s

running 10s outbound test (client to server) . . . . . 137.00 kb/s
running 10s inbound test (server to client) . . . . . . 470.22 kb/s

running 10s outbound test (client to server) . . . . . 141.00 kb/s
running 10s inbound test (server to client) . . . . . . 440.12 kb/s

running 10s outbound test (client to server) . . . . . 142.00 kb/s
running 10s inbound test (server to client) . . . . . . 311.90 kb/s

running 10s outbound test (client to server) . . . . . 138.00 kb/s
running 10s inbound test (server to client) . . . . . . 23.55 kb/s

running 10s outbound test (client to server) . . . . . 146.00 kb/s
running 10s inbound test (server to client) . . . . . . 609.52 kb/s

running 10s outbound test (client to server) . . . . . 143.00 kb/s
running 10s inbound test (server to client) . . . . . . 49.11 kb/s

running 10s outbound test (client to server) . . . . . 144.00 kb/s
running 10s inbound test (server to client) . . . . . . 113.20 kb/s

running 10s outbound test (client to server) . . . . . 139.00 kb/s
running 10s inbound test (server to client) . . . . . . 35.29 kb/s

running 10s outbound test (client to server) . . . . . 135.00 kb/s
running 10s inbound test (server to client) . . . . . . 640.87 kb/s

running 10s outbound test (client to server) . . . . . 138.00 kb/s
running 10s inbound test (server to client) . . . . . . 92.75 kb/s

running 10s outbound test (client to server) . . . . . 137.00 kb/s
running 10s inbound test (server to client) . . . . . . 570.63 kb/s

running 10s outbound test (client to server) . . . . . 145.00 kb/s
running 10s inbound test (server to client) . . . . . . 581.41 kb/s

running 10s outbound test (client to server) . . . . . 146.00 kb/s
running 10s inbound test (server to client) . . . . . . 107.77 kb/s

running 10s outbound test (client to server) . . . . . 144.00 kb/s
running 10s inbound test (server to client) . . . . . . 569.18 kb/s

running 10s outbound test (client to server) . . . . . 141.00 kb/s
running 10s inbound test (server to client) . . . . . . 451.00 kb/s

running 10s outbound test (client to server) . . . . . 142.00 kb/s
running 10s inbound test (server to client) . . . . . . 33.76 kb/s

running 10s outbound test (client to server) . . . . . 146.00 kb/s
running 10s inbound test (server to client) . . . . . . 219.28 kb/s

running 10s outbound test (client to server) . . . . . 140.00 kb/s
running 10s inbound test (server to client) . . . . . . 87.43 kb/s

running 10s outbound test (client to server) . . . . . 131.00 kb/s
running 10s inbound test (server to client) . . . . . . 536.87 kb/s

running 10s outbound test (client to server) . . . . . 136.00 kb/s
running 10s inbound test (server to client) . . . . . . 34.53 kb/s

running 10s outbound test (client to server) . . . . . 144.00 kb/s
running 10s inbound test (server to client) . . . . . . 486.64 kb/s

running 10s outbound test (client to server) . . . . . 136.00 kb/s
running 10s inbound test (server to client) . . . . . . 473.50 kb/s

running 10s outbound test (client to server) . . . . . 144.00 kb/s
running 10s inbound test (server to client) . . . . . . 414.87 kb/s

running 10s outbound test (client to server) . . . . . 137.00 kb/s
running 10s inbound test (server to client) . . . . . . 110.35 kb/s

running 10s outbound test (client to server) . . . . . 145.00 kb/s
running 10s inbound test (server to client) . . . . . . 90.96 kb/s

running 10s outbound test (client to server) . . . . . 142.00 kb/s
Connect() for Server to Client failed

running 10s outbound test (client to server) . . . . . 146.00 kb/s
running 10s inbound test (server to client) . . . . . . 563.02 kb/s

running 10s outbound test (client to server) . . . . . 147.00 kb/s
running 10s inbound test (server to client) . . . . . . 498.58 kb/s

running 10s outbound test (client to server) . . . . . 146.00 kb/s
running 10s inbound test (server to client) . . . . . . 327.90 kb/s

running 10s outbound test (client to server) . . . . . 128.00 kb/s
running 10s inbound test (server to client) . . . . . . 100.34 kb/s

running 10s outbound test (client to server) . . . . . 127.00 kb/s
running 10s inbound test (server to client) . . . . . . 102.67 kb/s

----------

Notice how the outbound (upload) speed was very consistent, and how variable the inbound (download) speed was. In particular, note how the inbound (download) speed was often less than the outbound (upload) speed.

Phil
Phil_C

Phil_C

Member

This may be a clue as to what is going on:
quote:
Information: Other network traffic is congesting the link

Every test I've run using web100clt contains that message.

I did a quick Google search using these keywords:
"at&t" (congested OR congestion) network DSL

The results:
www.google.com/search?q=%22at%26t%22+%28congested+OR+congestion%29+network+DSL

Two of the results that turned up:
»Why is AT&T Capping DSL Users, but Not U-Verse Users? [70] comments
»at&t dsl congestion in silicon valley?

Phil