Network Speed Test Help

[jbray]

What is good software to use to test my speed over my gigabit lan? I want to make sure my switches are still up to par.

 

[hang-the-9]

Try this one http://www.totusoft.com/lanspeed1.html

 

[jbray]

Try this one http://www.totusoft.com/lanspeed1.html

Thanks this is just what I was looking for. Simple yet effective.

 

[JOSHSKORN]

Sorry for bumping an old topic, but I found this from Google. I'm looking to achieve a gigabit connection. My speed, according to this test is 305Mbps. The cable I use to connect to my router, I've had for probably 6 years, I got basically a crossover cable, which I think in my case is a Cat5e. If I were to replace it with a Cat6, what am I to expect, speed wise? Mind you, my PC's CPU is a E6750 with a SATA II HDD. Not sure how much that comes into play. I also have Verizon FiOS, and have a LAN-TO-LAN setup (Actiontec to Asus router). I want Gigabit speed to my ASUS router (which I'm currently wired into.

 

[bill001g]

The cable will make no difference. Cat5e is designed to run 1g. Even if you were to buy cat7 cable it would run no faster. The speed of the data though the wire is the same it is based on some fraction of the speed of light based on using copper. Very technically it takes longer in cat7 cable because there is a tiny bit more wire because of the twists but only a very obsessive person would even attempt to calculate this because it can't be detected in actual use. The difference is not the speed of the media it is how much data you encode into the signal. This is controlled by the interfaces. If you had 10g interfaces better cable will allow you to use a different encoding method. But 1g interfaces will still use the 1g encoding method on even on better cable. It is the interface that controls the amount of data that is sent not the cable.

If you have 1g interfaces on your devices the data is ALWAYS transmitted at 1g/sec. The issue is how it is reported on end machines. So lets say i have 1gbit of data to send. It must be send in 1 second since that is the way the ports work and if they computer would look right at the end of that 1 sec it would see 1gbit/sec. But lets say it only looks every 10 sec now the rate is 100mbit/sec because I send data at 1gbit for 1 second and transmitted nothing for 9 seconds so my average time is only 100mbit/sec. In real machine obviously it check more often that 10 sec but you are also sending data chunks that are much smaller also. The key problem is why the machine would send data for 1 second and then send nothing for 9. This is almost always due to things like file structures or disk or something along that line.

The only way a cable would be the issue if you were getting errors which requires the data to be resent. Some nic cards will display the number of errors you get. It should be a very small number maybe 1 or 2 a week and that would be kinda high. So if you are getting errors on the ports a new cable might help. You no longer need cross cables when you use gig. Not sure if they hurt anything but don't spend extra for them since the concept does not even exist at gig speed.

There is a even simpler tool called IPERF. It is a old line mode tool. It runs completely from memory and is extremely tiny in size. It does not even use disk or much cpu so it try to eliminate delays caused by the end devices. It is doing raw data transfer without using any of the microsoft or other file protocols which slow things down. This tool is extremely basic and therefore is a good measure. You want both your pc you are testing to connected to lan ports, any wireless will greatly degrade your speed. You can likely get close to 900mbit/sec.