Need help tweaking

jhr76 · Post by **jhr76** » Fri Jun 08, 2007 10:03 pm

Hi. I recently changed my ISP. My TCP/IP is currently optimized (following Trogers advice) for a 640 k connection. My new connection has an advertised download speed of 2,5 Megs and an upload speed of 256 Kbs upload. I would like some help optimizing my TCP/IP (results from the TCP Analyzer and a couple of speed tests are included below). I would also like to know if I should increase the number of half open TCP connections (it is currently set to 100) and what speed should I limit my uploads to when using P2P software (mostly Bitcomet and Limewire). Here are the Analyzer results:
« SpeedGuide.net TCP Analyzer Results »
Tested on: 06.08.2007 21:44
IP address: 200.55.xxx.xx

TCP options string: 020405a001010402
MSS: 1440
MTU: 1480
TCP Window: 65535 (NOT multiple of MSS)
RWIN Scaling: 0
Unscaled RWIN : 65535
Reccomended RWINs: 63360, 126720, 253440, 506880
BDP limit (200ms): 2621kbps (328KBytes/s)
BDP limit (500ms): 1049kbps (131KBytes/s)
MTU Discovery: ON
TTL: 111
Timestamps: OFF
SACKs: ON
IP ToS: 00000000 (0)

Speed test from a local ISP (it only measured download speed):
First 128K took 531 ms = 246840 Bytes/sec = Approximate speed: 1975 Kbps
Second 128K took 484 ms = 270810 Bytes/sec = Approximate speed: 2166 Kbps
Third 128K took 500 ms = 262144 Bytes/sec = Approximate speed: 2097 Kbps
Fourth 128K took 485 ms = 270252 Bytes/sec = Approximate speed: 2162 Kbps
Results from Speedtest.net:
2174 kb/s download
113 kb/s upload
Latency 246 ms
Distance approx 550 miles (it was the recommended server located in Brazil, I am in Argentina)
Thanks a lot in advance.
Javier

trogers · Post by **trogers** » Fri Jun 08, 2007 11:27 pm

Try the following with TCP Optimizer:

General Settings tab:
Custom settings - check
Modify All Network Adapters - check
network adapter selection - your NIC
MTU 1480
TTL - 64
TCP Receive Window - 126720
MTU Discovery - Yes
Black Hole Detect - No
Selective Acks - Yes
Max Duplicate ACKs - 2
TCP 1323 Options:
Windows Scaling - checked
Timestamps - uncheck

Advanced Settings tab:
Max Connections per Server - 10
Max Connections per 1.0 Server - 20
LocalPriority - 1
Host Priority - 1
DNSPriority - 1
NetbtPriority - 1
Lan Browsing speedup - optimized
QoS: NonBestEffortLimit - 0
ToS: DisableUserTOSSetting - 0
ToS: DefaultTOSValue - 80
MaxNegativeCacheTtl - 0
NetFailureCacheTime - 0
NegativeSOACache Time - 0
LAN Request Buffer Size - 32768
Then select "Apply Changes" and reboot to take effect

After reboot, test your speed at this link using the recommended test server:

http://www.speedtest.net/

jhr76 · Post by **jhr76** » Sat Jun 09, 2007 8:20 am

Thanks man. I will try those tweaks, run another speed test and report back. BTW, should I leave the number of half open TCP connections in 100 or should I increase it for my new bandwidth? Thanks again

jhr76 · Post by **jhr76** » Sat Jun 09, 2007 8:35 am

Ok, I applied the tweaks and run a new speedtest with the recommended server. Here are the results:
2168 kb/s download speed
114 kb/s upload speed
Latency 242 ms
Distance approx 550 miles
A bit less latency than before, but the download speed was also a bit lower and the upload speed was 1 k higher. In general the download speed seems just fine for what I pay for (2,5 megs), though the upload speeds seems rather low for an advertised 256 ks, even though I have no point of comparison...
If you have any other ideas please let me know. Also please check the question in my previous post. Thanks again.

trogers · Post by **trogers** » Sat Jun 09, 2007 11:58 am

Try scanning with this new edition of Ad-Aware: https://www.speedguide.net/forums/viewtopic.php?t=223654

jhr76 · Post by **jhr76** » Sat Jun 09, 2007 7:18 pm

I am downloading right now, but my computer should be clean, I use Spy Sweeper 5.2, SpywareBlaster and NOD32, and of course Firefox, but I´ll give it a try. BTW, I have Windows XP Home, SP 2, fully updated. Once again, should I increase the number of half open TCP connections (currently set to 100)?. I have noticed the change in speed while using Limerwire and direct downloads, but I´m still getting the same speed I used to get with my 640 k connection while using Bitcomet. Thanks again

jhr76 · Post by **jhr76** » Sat Jun 09, 2007 10:26 pm

OK, I run Ad Aware 2007, it found some traces (6) and some negligible items. I also run Spy Sweeper and NOD32, so my computer is now as clean as it can possibly get. Here are the new results:
2163 kb/s download
110 kb/s upload
Latency 248 ms
same server as before
So even lower than before tweaking...

Here are the results from the local ISP:
First 128K took 485 ms = 270252 Bytes/sec = Approximate speed: 2162 Kbps
Second 128K took 1875 ms = 69905 Bytes/sec = Approximate speed: 559 Kbps
Third 128K took 484 ms = 270810 Bytes/sec = Approximate speed: 2166 Kbps
Fourth 128K took 500 ms = 262144 Bytes/sec = Approximate speed: 2097 Kbps

trogers · Post by **trogers** » Sat Jun 09, 2007 11:16 pm

Try this test and post the report under the 'Statistics' button:

http://nitro.ucsc.edu/

jhr76 · Post by **jhr76** » Sun Jun 10, 2007 8:11 am

OK, here are the results for this test (BTW, I kept my firewall, Sygate Personal Firewall, off during this and all the previous tests):
WEB100 Enabled Statistics:
Checking for Middleboxes . . . . . . . . . . . . . . . . . . Done
checking for firewalls . . . . . . . . . . . . . . . . . . . Done
running 10s outbound test (client-to-server [C2S]) . . . . . 194.0kb/s
running 10s inbound test (server-to-client [S2C]) . . . . . . 1.82Mb/s

------ Client System Details ------
OS data: Name = Windows XP, Architecture = x86, Version = 5.1
Java data: Vendor = Sun Microsystems Inc., Version = 1.5.0_09

------ Web100 Detailed Analysis ------
Cable modem/DSL/T1 link found.
Link set to Full Duplex mode
No network congestion discovered.
Good network cable(s) found
Normal duplex operation found.

Web100 reports the Round trip time = 448.14 msec; the Packet size = 1440 Bytes; and
No packet loss was observed.
C2S throughput test: Packet queuing detected: 3.16%
S2C throughput test: Packet queuing detected: 4.00%
This connection is receiver limited 72.38% of the time.
This connection is network limited 27.46% of the time.

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: OFF
RFC 1323 Window Scaling: ON

Server 'nitro.ucsc.edu' is not behind a firewall. [Connection to the ephemeral port was successful]
Client is not behind a firewall. [Connection to the ephemeral port was successful]
Information: Network Middlebox is modifying MSS variable
Server IP addresses are preserved End-to-End
Client IP addresses are preserved End-to-End

trogers · Post by **trogers** » Sun Jun 10, 2007 9:50 am

jhr76 wrote:OK, here are the results for this test (BTW, I kept my firewall, Sygate Personal Firewall, off during this and all the previous tests):
WEB100 Enabled Statistics:
Checking for Middleboxes . . . . . . . . . . . . . . . . . . Done
checking for firewalls . . . . . . . . . . . . . . . . . . . Done
running 10s outbound test (client-to-server [C2S]) . . . . . 194.0kb/s
running 10s inbound test (server-to-client [S2C]) . . . . . . 1.82Mb/s

------ Client System Details ------
OS data: Name = Windows XP, Architecture = x86, Version = 5.1
Java data: Vendor = Sun Microsystems Inc., Version = 1.5.0_09

------ Web100 Detailed Analysis ------
Cable modem/DSL/T1 link found.
Link set to Full Duplex mode
No network congestion discovered.
Good network cable(s) found
Normal duplex operation found.

Web100 reports the Round trip time = 448.14 msec; the Packet size = 1440 Bytes; and
No packet loss was observed.
C2S throughput test: Packet queuing detected: 3.16%
S2C throughput test: Packet queuing detected: 4.00%
This connection is receiver limited 72.38% of the time.
This connection is network limited 27.46% of the time.

Test shows your upload speed at 194 kbps to the US.

Test also shows packet queues due to slightly higher latency of 450ms compared with the 400ms that your TCP Window allows.

Try the following with TCP Optimizer:

General Settings tab:
Custom settings - check
Modify All Network Adapters - check
network adapter selection - your NIC
MTU 1480
TTL - 64
TCP Receive Window - 187200
MTU Discovery - Yes
Black Hole Detect - No
Selective Acks - Yes
Max Duplicate ACKs - 2
TCP 1323 Options:
Windows Scaling - checked
Timestamps - uncheck

Advanced Settings tab:
Max Connections per Server - 10
Max Connections per 1.0 Server - 20
LocalPriority - 1
Host Priority - 1
DNSPriority - 1
NetbtPriority - 1
Lan Browsing speedup - optimized
QoS: NonBestEffortLimit - 0
ToS: DisableUserTOSSetting - 0
ToS: DefaultTOSValue - 80
MaxNegativeCacheTtl - 0
NetFailureCacheTime - 0
NegativeSOACache Time - 0
LAN Request Buffer Size - 32768
Then select "Apply Changes" and reboot to take effect

jhr76 · Post by **jhr76** » Sun Jun 10, 2007 1:05 pm

OK. I made those changes but I´m not sure the TCP Optimizer is actually applying them because the confirmation window showed different values, so I am including a TCP Analyzer test, as well as another speed test and another test from that other site you told me about. Here they are:
« SpeedGuide.net TCP Analyzer Results »
Tested on: 06.10.2007 13:01
IP address: 200.55.xx.xxx

TCP options string: 020405a00103030201010402
MSS: 1440
MTU: 1480
TCP Window: 259200 (multiple of MSS)
RWIN Scaling: 2
Unscaled RWIN : 64800
Reccomended RWINs: 63360, 126720, 253440, 506880
BDP limit (200ms): 10368kbps (1296KBytes/s)
BDP limit (500ms): 4147kbps (518KBytes/s)
MTU Discovery: ON
TTL: 47
Timestamps: OFF
SACKs: ON
IP ToS: 00000000 (0)

WEB100 Enabled Statistics:
Checking for Middleboxes . . . . . . . . . . . . . . . . . . Done
checking for firewalls . . . . . . . . . . . . . . . . . . . Done
running 10s outbound test (client-to-server [C2S]) . . . . . 205.0kb/s
running 10s inbound test (server-to-client [S2C]) . . . . . . 1.38Mb/s

------ Client System Details ------
OS data: Name = Windows XP, Architecture = x86, Version = 5.1
Java data: Vendor = Sun Microsystems Inc., Version = 1.5.0_09

------ Web100 Detailed Analysis ------
Cable modem/DSL/T1 link found.
Link set to Full Duplex mode
Information: throughput is limited by other network traffic.
Good network cable(s) found
Normal duplex operation found.

Web100 reports the Round trip time = 336.72 msec; the Packet size = 1440 Bytes;

and
There were 47 packets retransmitted, 143 duplicate acks received, and 173 SACK

blocks received
The connection stalled 1 times due to packet loss
The connection was idle 0.59 seconds (5.89%) of the time
C2S throughput test: Packet queuing detected: 2.68%
S2C throughput test: Packet queuing detected: 3.12%
This connection is network limited 99.9% of the time.

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: OFF
RFC 1323 Window Scaling: ON

Server 'nitro.ucsc.edu' is not behind a firewall. [Connection to the ephemeral

port was successful]
Client is not behind a firewall. [Connection to the ephemeral port was

successful]
Information: Network Middlebox is modifying MSS variable
Server IP addresses are preserved End-to-End
Client IP addresses are preserved End-to-End

Speed test:
2175 kb/s
111 kb/s
247 ms latency
Recommended server
Once again thanks for all your help.

trogers · Post by **trogers** » Sun Jun 10, 2007 2:34 pm

Search Google, download, update and scan with the following programs:

SpyBot
Ad-Aware
a-Squared Free

jhr76 · Post by **jhr76** » Sun Jun 10, 2007 9:14 pm

I run all 3 programs (I had already run Ad Aware 2007) and they mostly found traces and tracking cookies. That a-Square program seems fishy to me, it identified as trojan the exe files for 2 of my games, Counter Strike and a Battlefield 2 mod which every other program I´ve run (Spy Sweeper, NOD32, Ad Aware and Spybot) considers safe, and it also considered malware a dll from my mouse drivers. I did not have it clean those. I run another test at Speedtest.net and my download speeds have now decreased dramatically even though the upload speeds did increase slightly. I am including those results as well as the one from a local ISP I included first (there wasn´t that much of a change there) and from that other test you had me run which also showed quite a decreased in speed/performance.
Speedtest.net results:
1328 kb/s dowload
190 kb/s upload
314 latency
Recommended server
Local ISP results:
First 128K took 484 ms = 270810 Bytes/sec = Approximate speed: 2166 Kbps
Second 128K took 1984 ms = 66065 Bytes/sec = Approximate speed: 529 Kbps
Third 128K took 500 ms = 262144 Bytes/sec = Approximate speed: 2097 Kbps
Fourth 128K took 485 ms = 270252 Bytes/sec = Approximate speed: 2162 Kbps

WEB100 Enabled Statistics:
Checking for Middleboxes . . . . . . . . . . . . . . . . . . Done
checking for firewalls . . . . . . . . . . . . . . . . . . . Done
running 10s outbound test (client-to-server [C2S]) . . . . . 205.0kb/s
running 10s inbound test (server-to-client [S2C]) . . . . . . 1.21Mb/s

------ Client System Details ------
OS data: Name = Windows XP, Architecture = x86, Version = 5.1
Java data: Vendor = Sun Microsystems Inc., Version = 1.5.0_09

------ Web100 Detailed Analysis ------
Cable modem/DSL/T1 link found.
Link set to Full Duplex mode
Information: throughput is limited by other network traffic.
Good network cable(s) found
Normal duplex operation found.

Web100 reports the Round trip time = 406.64 msec; the Packet size = 1440 Bytes; and
There were 164 packets retransmitted, 58 duplicate acks received, and 82 SACK blocks received
The connection stalled 1 times due to packet loss
The connection was idle 0.69 seconds (6.89%) of the time
C2S throughput test: Packet queuing detected: 2.84%
S2C throughput test: Packet queuing detected: 4.03%
This connection is network limited 99.82% of the time.

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: OFF
RFC 1323 Window Scaling: ON

Server 'nitro.ucsc.edu' is not behind a firewall. [Connection to the ephemeral port was successful]
Client is not behind a firewall. [Connection to the ephemeral port was successful]
Information: Network Middlebox is modifying MSS variable
Server IP addresses are preserved End-to-End
Client IP addresses are preserved End-to-End
Any other ideas?

trogers · Post by **trogers** » Mon Jun 11, 2007 12:16 pm

Search Google, download and run the free version of the PingPlotter to see which hops your packet losses are occurring.

jhr76 · Post by **jhr76** » Mon Jun 11, 2007 3:35 pm

Ok. How long should I run it for before posting my results and can I use other programs (such as Bitcomet) while it´s running?

trogers · Post by **trogers** » Mon Jun 11, 2007 9:54 pm

jhr76 wrote:Ok. How long should I run it for before posting my results and can I use other programs (such as Bitcomet) while it´s running?

Run it for 10 minutes will do. Avoid having any background internet traffic during testing.

jhr76 · Post by **jhr76** » Tue Jun 12, 2007 9:21 pm

Hi. Ok, here are the results. I don´t know if I did it right, there seems to be very little info. I did the trace/ping on my ISP´s page, http://www.flash.com.ar. I included the info in text format as well as a link for the image. Hope it helps.

Use this image link, it´s bigger: http://img529.imageshack.us/my.php?image=pingjk1.jpg
(I´m not used to uploading images)

Target Name: http://www.flash.com.ar
IP: 200.42.96.5
Date/Time: 12/06/2007 09:27:06 p.m. to 12/06/2007 09:34:21 p.m.

Hop Sent Err PL% Min Max Avg Host Name / [IP]
1 30 0 0,0 12 31 14 [10.10.10.2]
2 30 23 76,7 13 15 14 186.49.3.200.telecom.net.ar [200.3.49.186]
3 30 0 0,0 13 31 15 200-42-96-5.prima.net.ar [200.42.96.5]

http://img529.imageshack.us/img529/8773/pingjk1.th.jpg

trogers · Post by **trogers** » Wed Jun 13, 2007 7:08 am

Packet losses are occurring in hop 2, which is from the cable connection at your modem to the ISP's switch station.

First do a physical check for loose connection or loose/damaged wire of your cable line from your modem to the wall jack. If you have a splitter along this line, try a direct connection from the modem to the wall jack without the splitter to see if it solves your packet loss problem.

If the problem is not along this cable line, then you will need to call in your ISP tech to locate the source of your packet losses.

jhr76 · Post by **jhr76** » Wed Jun 13, 2007 5:57 pm

I intend to change the internal wiring within the next few days, which would allow me to do without the splitter I´m using now (also, even though it´s not directly related to this I´m thinking of formatting my PC and make a fresh start). If the problem persists I will contact my ISP. I suspect it´s a problem with the phone lines, and if it is I´m screwed... My previous ISP was owned by the phone company and they never changed the lines even though I had a claim to do so due to an issue with constant disconnections... Anyway, I´ll try. What should I tell the guys at the ISP exactly if the problem persists? Thanks for all your help

PS: once the wiring has been changed I will run Ping Plotter again and post my results. Thanks again.