Hi @NorthHill 

Thanks for posting in our business forum.

Simple as GW>>TL-SG2008P>>TL-SG3210XHP-M2?

I believe you should consider the setup like GW>>TL-SG3210XHP-M2>>TL-SG2008P. The 3210XHP is a layer better than 20008P.

And regarding the speed issue, take a look at the log if there is a problem with the port being blocked. That could be the reason why you have a slowed network.

If you see a log like that, consider the full examination of your network to rule out the loop.

Hi @NorthHill 

Thanks for posting in our business forum.

NorthHill wrote

@Clive_A I have checked the logs and I do not see any issues there.Any other ideas?

Slowness reflects in the speed. Is that correct?

Have you run iperf to test with two PCs between switches A and B to verify the line between them and the port speed can reach its 1Gbps limit(as one is a gigabit switch)?

If you can reach 1Gbps in this test, that means the connection is fine. The slowness resides in the LAN > WAN connection. Usually, the cause is the cabling, improper DNS servers set, and ISP.

Have you tested the direct connection from the router?

Hi @NorthHill 

Thanks for posting in our business forum.

NorthHill wrote

Yes, slow speed. Yes, I have done fairly extensive iperf testing. I did some testing yesterday. Anything connected to the SG3210XHP-M2 was getting speeds averaging 360-450mb.

 

Today, I attempted testing again. I should note that I have rebooted all TP-Link devices multiple times prior to the second attempt with iperf testing. For what it is worth, I have 1 gigbit symmetrical internet service, with speeds often exceeding that. I have used all brand new ethernet and OM4 cables for the following testing.

 

I have tested using server1 (local, directly connected to switch), server2 (local, directly connected to switch), and serverX (cloud based) via the SG2008P.

• Server1 > server2 averages 943mb
• Server2 > server1 averages 960mb
• Server2 > serverX averages 820mb

 

Next I tested via PC1 (2.5G NIC, local, directly connected to switch), server3 (10G SFP+, local, directly connected to switch), and serverX (same as above) via SG3210XHP-M2.

 

• PC1 > serverX averages 826mb PC1 > server3 averages 942mb
• server3 > PC1 averages 944mb
• Here is one odd one: Server3 > serverX 509mb is best I've seen. I tried dozens of test, averaging ~390 to 420mb.
• Finally, I tested PC1 (on SG3210XHP) > server1 (on SG2008P) and get 942mb. Reverse was identical.

 

So, I suspect two things now. It would appear that a reboot of all devices may have helped. It also may appear that the ER7206 may be part of the problem as that only has gigabit connection. The ER707-M2 that I purchased was DOA and I have requested a refund from the seller. So that testing will have to wait.

That's bad for the DOA 707-M2.

Regarding the test between the server 1 and 2, it at least proves the switch is not to blame.

Try to move a server to the ER7206 bypassing the SG2008. We need to locate the connection problem.

I don't think it is a problem with the router but so far it's hard to say as the local iperf shows a slower speed around 400-500.

Hi @NorthHill 

Thanks for posting in our business forum.

NorthHill wrote

@Clive_A Okay, after a little more digging, I found out that I had enabled the firewall (and forgotten that I had done so) on the server that was directly connected to ER7206. Once that was disabled, I was able to run iperf in both directions, across both switches and the ER7206, with near gigabit speeds (945-974mb/s).

 

I did not think it would be a problem, but since I have the firewall on the ER7206 enabled, I am going to attempt leaving the firwall on that server disabled (at least temporarily) to see if that was causing a bottleneck. Given that each of the two switches are connected to the ER7206, I am guessing that is what is preventing me from getting 2.5Gb speeds on the network. I suspect that something like the following is occuring:

 

server1 > SG2008 > ER7206 > SG3210XHP > PC1

 

Another scenario within the LAN looks like this:

 

PC2 > EAP660HD > SG3210XHP > ER7206 > SG3210XHP > server2

 

In the second scenario, even though PC2 is WiFi 6 and can transmit/receive speeds in excess of 1gb, and EAP660HD is connected via 2.5Gb on SG3210XHP, that the routing goes through ER7206 and is slowed because it only has 1Gb ports.

 

Last but not least, the server I mentioned above that had the firewall enabled also has my Omada software controller in a VM. So, my suspicion is that may have been slowing things down as well. I will wait and see.

Not sure why you are insisting on getting 2.5Gbps.

As for now, 900Mbps is the normal speed you should get between them. You have all gigabit devices but only a switch that's 2.5Gbps. Getting a gigabit speed is what you can expect from this unless your whole network is based on 2.5Gbps devices.