Apologies for the long post - I wanted to provide as much detail as I could. I've been having an issue for the last week and I can't seem to get to the bottom of it.
I have 3 switches in my network setup, interlinked as below:
TL-SG108 <--> HP Procurve 1400-8G <--> HP Procurve 1800-24G
There are multiple devices connected to the first 2 switches, but only a single device connected to the 1800-24G. It is a Dell R710 rack server. To rule out problems with link aggregation and failover, I've simplified it so that it has one gigabit NIC connected and the iDrac Enterprise 100mbps NIC connected (for remote management). The server is running ESXI (not sure if that matters). There are no VLANs configured.
I then have my desktop machine connected to the 1400-8G, running Windows 10.
I am using iperf to measure network throughput in both directions between these 2 machines. I get abysmal speeds of 1-4 MBps. Sometimes it is in both directions, sometimes one direction is better than the other. Occasionally, if I try different ports on the 1800-24G, it jumps up to the expected 112 MBps. However, it wont stay that way for long.
To rule out a faulty switch, I have tried plugging both machines into both the 1400-8G and the 1800-24G directly. Then I achieve a constant 90 - 112 MBps.
I know that inter-switch transfers shouldn't affect the speed so drastically, as I have another machine connected to the TL-SG108, which achieves max rates to my desktop consistently.
I'm thinking whether this article may shed some light:
https://www.smallnetbuilder.com/lanwan/lanwan-features/30212-when-flow-control-is-not-a-good-thing
The 100Mbps mgmt port on the R710 perhaps causing an issue? But this wouldnt explain why connecting directly to either switch is fine (mgmt port connected also).
I have 3 switches in my network setup, interlinked as below:
TL-SG108 <--> HP Procurve 1400-8G <--> HP Procurve 1800-24G
There are multiple devices connected to the first 2 switches, but only a single device connected to the 1800-24G. It is a Dell R710 rack server. To rule out problems with link aggregation and failover, I've simplified it so that it has one gigabit NIC connected and the iDrac Enterprise 100mbps NIC connected (for remote management). The server is running ESXI (not sure if that matters). There are no VLANs configured.
I then have my desktop machine connected to the 1400-8G, running Windows 10.
I am using iperf to measure network throughput in both directions between these 2 machines. I get abysmal speeds of 1-4 MBps. Sometimes it is in both directions, sometimes one direction is better than the other. Occasionally, if I try different ports on the 1800-24G, it jumps up to the expected 112 MBps. However, it wont stay that way for long.
To rule out a faulty switch, I have tried plugging both machines into both the 1400-8G and the 1800-24G directly. Then I achieve a constant 90 - 112 MBps.
I know that inter-switch transfers shouldn't affect the speed so drastically, as I have another machine connected to the TL-SG108, which achieves max rates to my desktop consistently.
I'm thinking whether this article may shed some light:
https://www.smallnetbuilder.com/lanwan/lanwan-features/30212-when-flow-control-is-not-a-good-thing
The 100Mbps mgmt port on the R710 perhaps causing an issue? But this wouldnt explain why connecting directly to either switch is fine (mgmt port connected also).