Mlab test optimal recieve bufffer?

General discussion related to Cable Modems, DSL, Wireless, Fiber, Mobile Networks, Wireless ISPs, Satellite, or any other type of high-speed Internet connection, general issues and questions here. Review and discuss ISPs as well (AT&T / SBC, BellSouth, Bright House, CableOne, Charter, Comcast, Covad, Cox, Cablevision / Optimum Online, TMobile, Verizon FIOS, Shaw, Telus, Starlink, etc.)
Post Reply
IIMaxII
New Member
Posts: 3
Joined: Fri Jul 06, 2012 3:47 pm

Mlab test optimal recieve bufffer?

Post by IIMaxII »

Your system: Windows 7 version 6.1
Java version: 1.7.0_05 (amd64)

TCP receive window: 66560 current, 66560 maximum
1.0E-6 packets lost during test
Round trip time: 31 msec (minimum), 119 msec (maximum), 104.66 msec (average)
Jitter: 88 msec
0 seconds spend waiting following a timeout
TCP time-out counter: 306
0 selective acknowledgement packets received

No duplex mismatch condition was detected.
The test did not detect a cable fault.
No network congestion was detected.
No network address translation appliance was detected.

0.0322% of the time was not spent in a receiver limited or sender limited state.
96.39% of the time the connection is limited by the client machine's receive buffer.
Optimal receive buffer: 68157440 bytes
0 duplicate ACKs set

WEB100 Kernel Variables: Client: localhost/127.0.0.1 CurMSS: 1452 X_Rcvbuf: 87380 X_Sndbuf: 175592 AckPktsIn: 2147 AckPktsOut: 0 BytesRetrans: 0 CongAvoid: 0 CongestionOverCount: 0 CongestionSignals: 0 CountRTT: 2147 CurCwnd: 68244 CurRTO: 306 CurRwinRcvd: 66560 CurRwinSent: 5888 CurSsthresh: 2147483647 DSACKDups: 0 DataBytesIn: 0 DataBytesOut: 6384476 DataPktsIn: 0 DataPktsOut: 4338 DupAcksIn: 0 ECNEnabled: 0 FastRetran: 0 MaxCwnd: 68244 MaxMSS: 1452 MaxRTO: 311 MaxRTT: 119 MaxRwinRcvd: 66560 MaxRwinSent: 5888 MaxSsthresh: 0 MinMSS: 1452 MinRTO: 231 MinRTT: 31 MinRwinRcvd: 43520 MinRwinSent: 5840 NagleEnabled: 1 OtherReductions: 0 PktsIn: 2147 PktsOut: 4338 PktsRetrans: 0 RcvWinScale: 7 SACKEnabled: 3 SACKsRcvd: 0 SendStall: 0 SlowStart: 45 SampleRTT: 105 SmoothedRTT: 106 SndWinScale: 8 SndLimTimeRwin: 9695719 SndLimTimeCwnd: 323552 SndLimTimeSender: 39112 SndLimTransRwin: 1 SndLimTransCwnd: 8 SndLimTransSender: 8 SndLimBytesRwin: 6189760 SndLimBytesCwnd: 164864 SndLimBytesSender: 29852 SubsequentTimeouts: 0 SumRTT: 224700 Timeouts: 0 TimestampsEnabled: 0 WinScaleRcvd: 8 WinScaleSent: 7 DupAcksOut: 0 StartTimeUsec: 612561 Duration: 10061492 c2sData: 2 c2sAck: 2 s2cData: 8 s2cAck: 3 half_duplex: 0 link: 10 congestion: 0 bad_cable: 0 mismatch: 0 spd: 5.08 bw: 105.85 loss: 0.000001000 avgrtt: 104.66 waitsec: 0.00 timesec: 10.00 order: 0.0000 rwintime: 0.9639 sendtime: 0.0039 cwndtime: 0.0322 rwin: 0.5078 swin: 1.3397 cwin: 0.5207 rttsec: 0.104658 Sndbuf: 175592 aspd: 0.00000 CWND-Limited: 328.20 minCWNDpeak: -1 maxCWNDpeak: -1 CWNDpeaks: -1 The theoretical network limit is 105.85 Mbps The NDT server has a 85.0 KByte buffer which limits the throughput to 12.80 Mbps Your PC/Workstation has a 65.0 KByte buffer which limits the throughput to 4.85 Mbps The network based flow control limits the throughput to 4.97 Mbps Client Data reports link is 'T1', Client Acks report link is 'T1' Server Data reports link is 'OC-48', Server Acks report link is 'Ethernet'


:confused:
How may I edit the receive buffer? Do they mean "receive buffer" in my nics advanced settings?
User avatar
Philip
SG VIP
Posts: 11733
Joined: Sat May 08, 1999 5:00 am
Location: Jacksonville, Florida

Post by Philip »

The "receive buffer" is your "TCP Receive Window" (RWIN) value.

Under Windows 7 the receive window cannot be directly set, it is being "auto-tuned" by the OS depending on your network. You can, however modify how aggressively the auto-tuning algorithm allocates the buffer, you may want to check the TCP Optimizer program. We also have an article on the main site for manually editing the settings.

I hope this helps.
drbaltazar
New Member
Posts: 6
Joined: Thu Aug 16, 2012 2:45 pm

Post by drbaltazar »

op:are you using web creator tool?(http://netalyzr.icsi.berkeley.edu/)if so you ll need to use a linux program ,since i dont know how that fix is implemented i wont try to explain ,suffice it to say the buffer you experience is at isp end.dont think you can do much aside from supplying the result netalyzr is gona give you.(by the way these guys are some of those that invented the web we know and love.dont sweat it they know whats what!for a fix ?only thing maybe remotly close is futur w8 netsh command dctcp (not enabled by default)
it basicly gona make use of you ecn result (if you activated it)it should eleviate the 1300 ms mega buffering everybody get!sorry i cannot be of more help,this beffering wont be notice unless you use a test like i supplied .(ya it is very hard to test!)(java only for this test)
again sorry ,dont forget to disable java after you re done (to stay secure)
Post Reply