Yes, it’s a Cisco enterprise switch, so I can see pretty much anything on it. It has a TDR feature as well, and I ran that this afternoon and it also shows some cabling issues. While I typically don’t trust the built in TDR, I think it’s probably correct in this case.
This is not unusual. I find that even Wifi7 is usually slower than a dedicated wired 1GbE connection because just about anything can trip up Wifi - you sitting in the way of a unobstructed Line of Sight, interference from neighbors, and so on.
While some of these issues can be mitigated with the help of site surveys, etc. anything that has to move a lot of data reliably should be using a wired / fiber connection.
That is a separate issue. I suggest using a cable-tester to ensure that the physical cabling is OK. Then move on to fixing the negotiation issues. I found, for example, that Mikrotik hardware doesn’t always play nice auto-negotiating a SFP+ connection with a TrueNAS.
FWIW, a 10GbE fiber connection to the NAS allows about 400MB/s to be transferred, as long as the file is large - NAS has one Z3 VDEV and one sVDEV.
I always prefer a wired connection over WiFi, but I would have expected better speeds since I have 3 hotspots throughout my house. I suspect they aren’t tuned as well as they could be - I don’t have a lot of experience in that area. I’m getting throughput nearing or at capacity of FE connectivity, so I hope when I am able to fix the cabling and run the link at gig speeds that the speeds will increase about 10-fold.
My TrueNAF has two bonded 1GB connections, so the throughput there should be enough.
For the Wifi, check what channels they are using. For 2.4GHz, only three are worthwhile, i.e. 1,6,13 due to overlap. At 5GHz, there are more channels, I would similarly hard-code them to be far apart and widen them, unless there is local interference.
Wired LAGGs (if that’s what you are using) can typically double the throughput of the total network trunk, but not for individual users who will still be limited to 1GbE or whatever the capacity of each individual LAGG element is.
I never bank on fast WiFi speeds, not to compare to wired connections. WiFi is nice but if you are trying to transfer a lot of data fast, this isn’t the way unless it is the only connectivity you have, and then just accept it will be slow.
@Constantin made some very good points about hard coding the bands and which ones to use. It can make all the difference in the world.
I’m in the US, so I’m using 1,6,11, which I believe are the best channels to not overlap frequencies. I don’t have an option for channel 12 or 13, and I believe that is region based.
I am using LAGGs and understand that any bottleneck between the server and the client is the fastest speed the link is capable of, but the LAGGs could provide additional throughput (up to a max of the LAGG speed) for additional clients.
I’ve been playing around with channel width and transmit power, but not having much success by way of increased performance. I’ll keep at it, and worst case I’ll probably learn more about it.
Apologies, you are correct, it’s 1,6, and 11 in the US for 2.4GHz @ 20MHz each. For 5GHz, I use 80 MHz-wide channels, ditto for 6GHz. The key is keeping the channels from overlapping. What I still do not comprehend is the low wired speed, that suggests either broken / misconfigured wire pairs or a bad auto-negotiation.
Low wired speeds are because I have an autonegotiation issue caused by what appeared to be faulty wiring between my client and my switch. I thought I needed to reterminate the cabling in the wall, because that is where the problem appeared to be. Running the TDR function on my switch confirmed that could be the problem. While I don’t normally trust the TDR function, I thought maybe it was correct this time.
I was wrong. After checking the wiring, everything looked ok. I ordered a USB-C ethernet adapter for my mac mini. That works perfectly at 1Gbps. I’m really glad I paid extra for a 10GB ethernet port in my mac mini that doesn’t work. I now see that others have had similar autonegotiation issues with the 10GB adapters.
[ 7] local 172.16.110.29 port 63398 connected to 172.16.110.15 port 5201
[ ID] Interval Transfer Bitrate
[ 7] 0.00-1.00 sec 89.2 MBytes 749 Mbits/sec
[ 7] 1.00-2.00 sec 113 MBytes 942 Mbits/sec
[ 7] 2.00-3.00 sec 112 MBytes 942 Mbits/sec
[ 7] 3.00-4.00 sec 112 MBytes 942 Mbits/sec
[ 7] 4.00-5.00 sec 112 MBytes 942 Mbits/sec
[ 7] 5.00-6.00 sec 113 MBytes 942 Mbits/sec
[ 7] 6.00-7.00 sec 112 MBytes 942 Mbits/sec
[ 7] 7.00-8.00 sec 112 MBytes 941 Mbits/sec
[ 7] 8.00-9.00 sec 112 MBytes 942 Mbits/sec
[ 7] 9.00-10.00 sec 112 MBytes 942 Mbits/sec
- - - - - - - - - - - - - - - - - - - - - - - - -
[ ID] Interval Transfer Bitrate
[ 7] 0.00-10.00 sec 1.07 GBytes 922 Mbits/sec sender
[ 7] 0.00-10.01 sec 1.07 GBytes 922 Mbits/sec receiver
iperf Done.
Aquantia strikes back?
I really like the QNAP Thunderbolt 3 to SFP+ adapter. Bus powered and I can use any SFP+ transceiver in it.
Are you using a 10Gib port on the client to connect to a 1Gib port on the switch?
Does all the cabling meet 10G specs?
Is the cable length within the limits of 10G?
I know you are expecting it to downspeed to 1G (some are rated to downspeed and others are not), but the physical circuit in the computer is still a 10G circuit and expecting 10G spec cabling.
You mention cabling in the wall, every time you go through a connector you introduce an impedance bump which causes increased error rates. I have always used SFP+ for 10G with either SFP+/SFP+ cables or fiber. And you still need to watch switch port error rates and replace bad / damaged cables (much moire common than with 1G).
Yes, the 10Gb is on the client, and the switch is 1Gb, but the client should be backwards compatible with 1000/100/10 Mbps connections. My new USB-C adapter is 2.5 GBPs capable, but negotiates to 1Gb to match the switch port. The cabling is 10GB rated, but I only have a couple 10Gb ports and I have those connected to a couple servers.
My transfer rates and response times seem much more normal with the new network adapter , so I think that was the root of my problems.
Also, some 10G NICs have problem auto negotiating 1g speed, maybe you need to force it on both the PC and the router side to 1G (If your switch is managed, or at least a smart switch)
Is it still under warranty, if so, have them fix it? Not sure if the studios are the same, but my M1 studio does 10gbps to my Fiberstore switch just fine using the built in 10g nic.
This has also been my experience, especially in the context of MikroTik switches.
For whatever reason, the best option is to manually program network hardware speed / duplex / flow control settings for TrueNAS-MikroTik connections rather than rely on auto negotiation if both ends are not 10GbE.
For whatever reason I never had problems with 10GbE connections but constantly ran into down connections at 1GbE if auto-negotiation was used. I never hunted down if it was MikroTik or TrueNAS at fault here but suspect it was MikroTik.
So I would try out turning off auto-neg and see if hard-coded network settings do the trick. Remember to do it on both ends of the connection.