What's the problem in my setup?

[qflyer]

Hoping someone has seen something like this and can tell me what the fix is. I'm stumped.

There are a few more devices connected than I'm mentioning, but I'm not sure they're important. Basically, I have Starlink internet (30-200 Mbps depending on time of day, weather, etc) ---> netgear router ---> switch.

On that first switch, I have a wireless TP-Link point to point bridge which links with another TP Link bridge on my boat dock. That bridge feeds another switch. I have 4 cameras on that switch and also a Linksys wifi router configured as an access point for wifi access on the dock.

Bottlenecks:
Both switches are 7 port, all of which are 10/100 Mbps. MokerLink 7 Port PoE Switch with 5 PoE+ Port

Both TP-Link bridges are advertised as 300 Mbps but only have a 100 Mbps ethernet port, effectively making them 100 Mbps units even though it does establish a 300 Mbps wireless link between the units.

5GHz 300Mbps 13dBi Outdoor CPE

5GHz 300Mbps 13dBi Outdoor CPE

www.tp-link.com

The issue: with the dock cameras running, the switch is handling 45 Mbps. The bridges are handling that 45 Mbps as well with no problems. When I connect a laptop to the linksys router (or switch, doesn't matter) I can only get 3-7 Mbps on internet speed tests.

I've run iPerf with my laptop on the boat dock router (wireless or wired, makes no difference) and I get 45-50 Mbps throughput to my PC which is on the netgear router at the house. If I turn off the cameras, I get 95 Mbps through iPerf. So the switches and wireless bridges are perfectly capable of running at roughly their max speed of 100 Mbps.

So why can I only get 3-7 Mbps on internet speed tests at the dock? I think the switch is the culprit, but don't know what I'm missing. If iPerf can get 95 Mbps through it, shouldn't I be able to get internet at that speed? Yes, I've tested during off hours when I'm certain starlink is fast, at least 100+ Mbps...

 

[jmhmcse]

the two utilities while overlap in purpose/function don provide a 1-for-1 like comparable results

iperf is designed to be a hi-speed, force maximum throughout, find bottleneck network segments; it attempts to consume all the bandwidth at the expense of everything/everyone else on the network
speedtest is a general purpose, what is typical/standard available last-mile (or space-link) connection to the internet; plays nice with everyone else on network


possible points to ponder...

QOS settings on netgear router
"a few more devcies"; yes, they do matter... these are only 100Mbs switches

• router, besides switch how many other devices are connected
• switch, besides router and bridge how many other devices are connected

which ports of mokerlink switch are being used to connect bridging devices at each side of link; port 1 and and 2 are QOS'ed
bridging link

• signal quality us excellent between house/boat dock
• configuration settings are 'fixed' rather than 'auto'
• static ip addresses
• is being monitoring, usage/performance, while testing from boat dock

netgear router is providing dhcp through all links/legs of network; i.e. linksys router has dhcp disabled
all wired connections into switches/routers are full duplex
which port of mokerlink is laptop connected, when wired, at the boat dock
when wired at the boat dock, were all the camera (and other devices) disabled when receiving the 3-7Mbps result

do your iperf command parameters define appropriate protocol, packet size, window size, upload/download, etc to mimic speedtest

have you used a private or public www iperf server? testing both sides of the bridge to see those results;

https://iperf.fr/iperf-servers.php

= = = = = =
the max internal rating of the mokerlink switch is identified as 1.4Gbps. this would be the minimum for a 7 port 100Mbps full duplex switch; 7x100x2 = 1400Mbps or 1.4Gpbs. the limiting factor of the switch is the 100Mbps uplink.

the 300Mbps bridge shouldn't be a bottleneck for the mokerlink switches; even at 100Mbs full duplex

netgear connection to mokerlink should be either of the QOS ports (port 1 or 2) and tp-link connection to mokerlink should be the other QOS port.

linksys connection to mokerlink should be either of the QOS ports

 

[qflyer]

Thanks so much! Awesome reply with good info and I think we're getting somewhere!

Switch one has the netgear router and TP-link bridge connected to it, nothing else. Neither were plugged into the QOS ports, but they are now. I'll be adding a few cameras to this switch in a few days once I have all the cables ran where they need to be, but for now, nothing else is on this switch.

Switch two on the boat dock has TP link bridge, Linksys router (AP mode), and 4 cameras. Router and TP-link were not in QOS ports, and one (don't remember which) was plugged into an uplink port.

So with the TP-link bridges in QOS ports, Netgear router in QOS, and the Linksys in an uplink port, I'm getting 20-25 Mbps internet speed tests at the boat dock plugged into the Linksys! I'll have to test later tonight, as Starlink is slow during peak hours - I was only getting 25-30 at the house before I went to mess with the dock connections.

I don't have enough POE ports to have both the TP-Link bridge and the Linksys on QOS - that leaves 3 POE ports and I need 4 for the cameras. I did leave one cam unplugged and put the bridge and linksys on QOS with the same 20 Mbps results. Unexpected result: iPerf speeds between the dock laptop and home PC bounce between 3-7 Mbps now!?!?

Should I run the TP-link bridge into the Linksys first and then feed the switch, or does it matter?

Why would internet speed tests pick up, but iPerf start performing slowly like internet speed tests had previously done?

I guess my next step is to leave it as is since everything seems to work like I want it to now, other than the weird iPerf results, or to just move on and replace the switches with their bigger brother, the 11 port gigabit PoE switch: MokerLink 11 Port Gigabit PoE Switch with 8 PoE+ Port, 2 Gigabit Uplink, 1 SFP Port

Any other combinations I should try?

I'm not sure about the IPerf servers you mentioned. I used the download .exe file on both PCs for my tests, but if there's another way to test, I'll give it a shot if I can find a how-to on youtube! Sorry I'm somewhat computer illiterate.

 

[jmhmcse]

nice to hear that there was improvement.

as for the boat dock, i'd connect the tp-link to one of the QOS ports and linksys into one of the non-poe 'up-link' ports. perhaps not ideal, but should work. (strange that for a small 7, not even 8, port switch that all ports do not have poe)

the mokerlink switch you link to has a sfp port, do you intend to use this? similarly, does the uplink ports provide you with any benefit?

a more traditional eight port 1gbps switch like the mokerlink poe-g080g would be slightly cheaper alternative to the poe-g083gs; the g080g is $60 on amazon. one port for the tp-link bridge with seven additional ports for cameras and things.

to me the qos ports and uplink ports are gimmicky. on one of the mokerlink slides for their switch it appeared as though the uplink was a trunk-line; all traffic from all other ports was dumped to those ports all the time, not typically useful unless you're doing some packet sniffing with wireshark.

 

[qflyer]

nice to hear that there was improvement.

as for the boat dock, i'd connect the tp-link to one of the QOS ports and linksys into one of the non-poe 'up-link' ports. perhaps not ideal, but should work. (strange that for a small 7, not even 8, port switch that all ports do not have poe)

the mokerlink switch you link to has a sfp port, do you intend to use this? similarly, does the uplink ports provide you with any benefit?

a more traditional eight port 1gbps switch like the mokerlink poe-g080g would be slightly cheaper alternative to the poe-g083gs; the g080g is $60 on amazon. one port for the tp-link bridge with seven additional ports for cameras and things.

to me the qos ports and uplink ports are gimmicky. on one of the mokerlink slides for their switch it appeared as though the uplink was a trunk-line; all traffic from all other ports was dumped to those ports all the time, not typically useful unless you're doing some packet sniffing with wireshark.

Well, I can't say that I fully understand why it's behaving the way it does, but it's working well enough. Went down after peak hours so Starlink was consistently 100+ Mbps on the home PC. Tried everything as I left it earlier today with TP-Link on the QoS, Linksys on the uplink and got 20 Mbps consistently on the internet speed tests. iPerf was still 3-7 Mbps.

I changed it up just to see what would happen with the TP-Link going straight to the Linksys, then feeding the switch from Linksys. Internet speed tests stayed 20ish Mbps BUT iPerf went back to 45-50 Mbps. This was with all cameras running (45 Mbps), so total throughput with iPerf was around 90 Mbps, and total throughput with cameras plus speed tests was 65ish Mbps.

I had my big PTZ running at 20,480 Kbps, so I backed it off to 10,240, and backed my 5442 down a couple Mbps as well. That freed up about 12 Mbps. Internet speed tests increased by 5-6 Mbps, and iPerf increased the full 10-12 Mbps.

I think the switch is doing the best it can - maybe it's the limiting factor, or maybe there's a setting in the wireless bridges that I need to adjust to get better internet speed, but at least I have functional internet on the dock now. 25 Mbps is plenty for anything I need to do down there.

The bridges have perfect line of sight, roughly 400' apart. Here's some of its stats:
Signal strength: -38 dBm
Noise Strength: -90 dBm
SNR: 52 dB
TX and RX rate: 300 Mbps