Best way to test network speed?

[krowhill]

So I have a project that I am testing both for school and for personal and possibly business reasons in the future.  The project is to apply a low cost business grade computer as a router/firewall/DCPC and DNS server.  I have pfSense setup on a Dell 990.  It has an i5 processor, 8GB RAM, and two gigabit ethernet ports (one aftermarket one I added) and a 60GB HDD.  I am trying to test transfer speeds between the source computer (either a all in one micro Linux box with Kali Linux on it (Atom (I think) processor, 4GB RAM, live boot from thumb drive (no HDD on board), gigabit ethernet, or a MacBook Pro (i7, 16GB RAM, 1TB SSD, gigabit ethernet)) and the destination (Dell 9010 i7, 32GB RAM, 250GB SSD, (2) gigabit NICs (one for the test network the other for home network) with Windows 7, and Linux Mint dual boot).  On the MacBook I will have a 8GB RAM drive, and on the 9010 it will also have an 8GB RAM drive to eliminate possible bottleneck of read/write speeds of the SSD on source and destination.  Now here is the question, I want to see max speed throughput through the router/firewall/DHCP/DNS setup on the 990.  What would be the best method of finding the bottleneck?  So far I have tested using FTP and SCP, what other protocols or methods should I use to find the bottleneck?  I need to have a screen output of the total time spent on a transfer (so I can get a screen shot or take a photo of it if I can't screen shot).  Any thoughts?

[Mark77]

I find that iSCSI does a great job of saturating Gig-E. 

[Lurick]

iPerf, it's free and runs on Linux, Windows, and I believe Mac as well.

[krowhill]

1 minute ago, Lurick said:

iPerf, it's free and runs on Linux, Windows, and I believe Mac as well.

Does iPerf write from/to disks?

[Lurick]

2 minutes ago, krowhill said:

Does iPerf write from/to disks?

It can if you want it to test disks but by default it uses the memory instead since that's much faster than most equipment can handle it will give you a max throughput for network performance.

[krowhill]

5 minutes ago, Lurick said:

It can if you want it to test disks but by default it uses the memory instead since that's much faster than most equipment can handle it will give you a max throughput for network performance.

So if I understand you correctly I do not need to create RAM disks to test speed?

[Lurick]

1 minute ago, krowhill said:

So if I understand you correctly I do not need to create RAM disks to test speed?

Nope, it generates everything itself on memory, tests, and then removes it. Very simple to use and setup.

[krowhill]

Just now, Lurick said:

Nope, it generates everything itself on memory, tests, and then removes it. Very simple to use and setup.

Ok, I will test it and let you know how it goes!