Internet Speed Questions

[maxama10]

I just did a test for curiositys sake and the results surprised me. I have satellite internet through Hughes network and these speeds are about what is suggested.

I was only sitting about 5 ft away from my netgear 105Mbps router and used a Cat5e Belkin patch cable and then used the Apple built in wifi connection. Another thing, I live in rural Nevada and my internet is from somewhere back east but I used servers closest to my house, but once used the KY one as reccomended by speedtest.net.

Results:
Speedtest.net
SLC Wire
337 161
402 235 Avg= 373 / 215
380 250
Kentucky Wire
428 228
394 241 Avg= 411 / 234.5





SLC Wireless
585 233
452 241 Avg=580.6 / 229.6
705 215
Kentucky Wireless
650 199 Avg=654.5 / 209.5
659 220


Speakeasy.net
SanFrancisco
Wire
464 204 Avg=443 / 214
422 224

Wireless Avg=577.5 / 190.5
620 137
535 244


The left numbes are download speed in Kbps and the right numbers are upload speed in Kbps.

What I found odd is that the wireless speeds are all faster than the patch cable speeds. I had thought itd be the opposite, whats the deal? I always thought that cables were much faster. The only thing I can think of is that the cable was longer than it needed to be. Should it have that much of an affect on the speeds though? Another thing, it only seemes to affect download speeds, the upload speeds stayed pretty close regardless of wire/wireless.

Let me know what ya tthink?

[Hexis]

WiFi vs Wired = both are significantly faster then your Internet connection, so they are unrelated. As for the differences, there are many possible explainations. You do not have a cable that too long or anything. 100baseT is good to 328 feet over cat5. The signal propigation delay over even a 300 foot cable is not significant. There is no signal dedrigation over longer cables. In reality there is no real difference.

As for those "test" sites, I find them to be very inaccurate. I worked for an ISP for a few years. They always showed some tony fraction of the actual speeds we were able to push over a line with something simple like a nice FTP transfer.

There are also some TCP stack tweaking techniques to increase performance. If you set an appropiate window size, you may be able to get much better performance.

[maxama10]

There are also some TCP stack tweaking techniques to increase performance. If you set an appropiate window size, you may be able to get much better performance.


Im pretty technical minded when it comes to computers but you lost me here :D. How would I go about doing this?

[teufelhunden]

Im pretty technical minded when it comes to computers but you lost me here :D. How would I go about doing this?


http://www.dslreports.com/faq/tweaks

have at it :) make sure you download drtcp

[maxama10]

Im not seeing a download for OSX :confused:

[Hexis]

That's a great FAQ. The Window portion is here: http://www.dslreports.com/faq/903

TCP is a reliable protocol. It adds some overhead to provide that reliability. You minimize the overhead by setting an appropiate window size.

[Hexis]

http://slaptijack.com/system-administration/mac-os-x-tcp-performance-tuning/

You can look for BSD TCP tuning techniques. They will work on OSX as well.