Info Testing 4 popular Gigabit networking switches - Results are interesting!

[alexb75]

So, 3 weeks ago, I asked you guys if different gigabit switches would perform differently based on their spec or manufacturer. Consensus was, it doesn't and all will perform the same!

Being locked down with COVID, I went about testing that (too much time on my hand I guess)... so bought FOUR popular 5-port switches and tested them the following ways:

• Use LAN Test app to test Write/Read between two devices
• Connected both my PC and Laptop with ethernet cable to the switches
• Connected the router for internet access
• Connected my audio streamer

I ran 3x times of most tests to get an overall better result. I tested sending 500mb, 1GB, 3GB files... plus for the 3GB, I added playing 1-6 4K videos at the same time on both devices. Results are below:

Takeaways:

• They all performed great in easy scenarios, not much difference at all
• TP-Link has a massive drop off when you play videos at the same time! It's the only managed switch here, I tried changing QoS settings to see if it made a difference but it didn't'
• Not sure why TP-link which on paper has the best specs (more memory, more jumbo frames, managed, many more features), would perform so poorly. My suspicion is the QoS is messing things up, or their algorithm for switching isn't as good
• Netgear GS105, which is the old tested and true switch, with mediocre specs (even compared to its own cousin GS305), seemed to do best... BUT the top three performed so close, any would do
• My winner is GS105, as it was flawless everywhere, plus it has lifetime warrant (TP-link too but it didn't do as well) but costs $10 more

Just a COVID nerd thing to do and thought to share the results with you guys in case anyone is in the market, OR if some network engineer could explain the differences specially with TP-link!

Cheers and Happy Holidays!

Totusoft | LAN Speed Test

iPerf Chart

 

[kanewolf]

So, 3 weeks ago, I asked you guys if different gigabit switches would perform differently based on their spec or manufacturer. Consensus was, it doesn't and all will perform the same!

Being locked down with COVID, I went about testing that (too much time on my hand I guess)... so bought FOUR popular 5-port switches and tested them the following ways:

• Use LAN Test app to test Write/Read between two devices
• Connected both my PC and Laptop with ethernet cable to the switches
• Connected the router for internet access
• Connected my audio streamer

I ran 3x times of most tests to get an overall better result. I tested sending 500mb, 1GB, 3GB files... plus for the 3GB, I added playing 1-6 4K videos at the same time on both devices. Results are below:

Takeaways:

• They all performed great in easy scenarios, not much difference at all
• TP-Link has a massive drop off when you play videos at the same time! It's the only managed switch here, I tried changing QoS settings to see if it made a difference but it didn't'
• Not sure why TP-link which on paper has the best specs (more memory, more jumbo frames, managed, many more features), would perform so poorly. My suspicion is the QoS is messing things up, or their algorithm for switching isn't as good
• Netgear GS105, which is the old tested and true switch, with mediocre specs (even compared to its own cousin GS305), seemed to do best... BUT the top three performed so close, any would do
• My winner is GS105, as it was flawless everywhere, plus it has lifetime warrant (TP-link too but it didn't do as well) but costs $10 more

Just a COVID nerd thing to do and thought to share the results with you guys in case anyone is in the market, OR if some network engineer could explain the differences specially with TP-link!

Cheers and Happy Holidays!

To test network hardware you should be using iPerf, IMO. https://iperf.fr/
That removes most of the variables of the host end points from the test.

Was that the Helios LanTest tool you were using ? It creates files which puts additional dependencies on the host performance.

 

[alexb75]

To test network hardware you should be using iPerf, IMO. https://iperf.fr/
That removes most of the variables of the host end points from the test.

Was that the Helios LanTest tool you were using ? It creates files which puts additional dependencies on the host performance.

I used this Software Totusoft | LAN Speed Test

 

[kanewolf]

I used this Software Totusoft | LAN Speed Test

That also looks like it creates files. Have you tried a RAMDISK on each end for testing ?

 

[alexb75]

That also looks like it creates files. Have you tried a RAMDISK on each end for testing ?

No... but I could I guess. Since all other variables remained the same, the test should be valid. Specially since I did it multiple times (didn't record all in this chart), and 100% TP-link would bog down when I am playing 4K videos while transferring files.

I assume the slower write speed is due to SSD limit (Samsung 970 Pro NVME on the laptop side) but can't be sure.

 

[alexb75]

To test network hardware you should be using iPerf, IMO. https://iperf.fr/
That removes most of the variables of the host end points from the test.

Was that the Helios LanTest tool you were using ? It creates files which puts additional dependencies on the host performance.

Thanks for the recommendation, I just ran it with iPerf and it's much more reliable, agreed! Takes the file performance out of the way! Here are the results for iPerf! While send/receive are now consistent, the TP-link once again has the biggest drop-off when I play 4K videos! It's quite a massive hit... which again makes me wonder if they're QoS implementation is faulty! But even without running iPerf, TP-Link was the one that struggled the most just playing the video, and most were dropping off res, or getting interrupted. GS105 still did the best job, but the other two very close!

I am also posting the detailed iPerf results, you can see most maintain a similar speed from 1 second to next, TP-Link is very inconsistent as videos are being played, while it has the best performance if nothing else is being routed!

NetGear GS105

D-Link

Netgear GS305

TP-Link

 

[alexb75]

PS. Netgear GS105/305 look identical, apart from more LEDs for 1Gbps connection and the color of the chassis, BUT 105 takes 12V power while 305 takes 5V and GS105 is a bit heavier. So, I assume the chipset is different inside as every other spec seems identical (memory, buffer, etc)...