There is alot of discussion about using Jumbo Frames in a home environment and let me come out and say no I do not recommend it. However; most of us are looking to test it or use it in a lab and that is what I have done.
2 PC’s running Windows 8 with Intel 82579V Gigabit
The test: Do jumbo frames provide a significant boost in networkspeed and traffic. Please keep in mind that everyones network is different this was tested on a closed loop. I used some software called iperf to transfer data over 10 seconds 3 times. I only recorded the highs and the lows. In times where there was significant loss the test allowed for one run to be excluded when run again.
Also these images are for ants. I dont know why they are too small but Im too lazy to fix it.
Test 1 – Check normal transfer rates.
Result was normal and expected.
Test 3 – Check what happens when there are 2 hops that have jumbo frames enabled.
Results showed there was about 10MB loss per hop this is guesstimated that other hops would be the same as I had no other switching devices to add.
Test 4 – Run Jumbo Frames (4K) in an environment where some switching devices will not accept the frame.
Result showed that there was a significant loss in speed as the data becomes fragmented and resent.
Test 5 – 9K Frames baby!
This is where it got odd form me. I seemed to have a high loss in speed when I set 9K frames over my devices. I ensured all my hardware supported it however I can not explain why this happened. I will have to look into it later.
Test 6 – Jumbo Frames mixed MTU enviroment. This basically means that one PC is set to 4K and the other is set to 9K.
Test 7 – One sided Jumbo Transfer. In this layout only one PC and the switch have frames enabled.