Advice on iperf3 test results

[freddyq]

This takes me back to my first post on this topic...

Old machines and USB adapters aren't my favorite!

This takes me back to my first post on this topic...

Old machines and USB adapters aren't my favorite!

It's true, very frustrating. But does this effectively rule out an issue with the cables? At least until I can properly test them with the right hardware...

 

[DavidR1]

It's true, very frustrating. But does this effectively rule out an issue with the cables? At least until I can properly test them with the right hardware...

Rule out? No, unfortunately. It only shows that any test results from those cables are inconclusive.

But like you said, you can stop chasing your tail for now at least. Your "control" has to be 100% before any observations are relevant.

 

[freddyq]

Rule out? No, unfortunately. It only shows that any test results from those cables are inconclusive.

But like you said, you can stop chasing your tail for now at least. Your "control" has to be 100% before any observations are relevant.

OK yes - apart from trying to get my hands on two relatively new laptops with ethernet ports - do I have any other options?

 

[DavidR1]

OK yes - apart from trying to get my hands on two relatively new laptops with ethernet ports - do I have any other options?

Options as I see them:

• Borrow good hardware [work, friends, family, neighbor, student]. (Doesn't have to be a laptop though it would be easier to transport and setup of course.) Continue your tests.
• Be "that guy"* and "borrow" from an evil big box store you don't like with a generous return policy. Continue your tests. Return when done. *(Don't be that guy.)
• Buy the stuff you planned to buy anyway (PC for Blue Iris, cameras) and hope it all Just Works™. Figure it out afterwards if there actually are underlying issues.
• Pay someone to come terminate and certify the existing cable runs, with the caveat that there could actually be one or more problem runs and you'll have wasted money testing those with no way to fix them (i.e. hidden damage behind finished surfaces).
• Pay someone to install and certify all new cabling end-to-end with a guarantee so you never have to think about this again.
• Bribe someone [beer] local to you that can help (with hardware or test equipment) to validate what you've got at a fraction of the cost of a certified installation. Maybe there are clubs or other public resources available to you that might be able to help out (with either hardware, tools, or skills?)
• See if you or someone else comes up with any ideas we haven't already tried.

If you intend to move forward with your install eventually anyway (even if it means running new wires during new construction), it seems to make the most sense to just buy stuff and see what happens. Maybe nothing is wrong at all with the installed cabling and you've just gotten bad data due to an old machine and/or crappy USB adapter. Most other choices are economically foolish, unlikely, or just wrong.

 

[freddyq]

Options as I see them:

• Borrow good hardware [work, friends, family, neighbor, student]. (Doesn't have to be a laptop though it would be easier to transport and setup of course.) Continue your tests.
• Be "that guy"* and "borrow" from an evil big box store you don't like with a generous return policy. Continue your tests. Return when done. *(Don't be that guy.)
• Buy the stuff you planned to buy anyway (PC for Blue Iris, cameras) and hope it all Just Works™. Figure it out afterwards if there actually are underlying issues.
• Pay someone to come terminate and certify the existing cable runs, with the caveat that there could actually be one or more problem runs and you'll have wasted money testing those with no way to fix them (i.e. hidden damage behind finished surfaces).
• Pay someone to install and certify all new cabling end-to-end with a guarantee so you never have to think about this again.
• Bribe someone [beer] local to you that can help (with hardware or test equipment) to validate what you've got at a fraction of the cost of a certified installation. Maybe there are clubs or other public resources available to you that might be able to help out (with either hardware, tools, or skills?)
• See if you or someone else comes up with any ideas we haven't already tried.

If you intend to move forward with your install eventually anyway (even if it means running new wires during new construction), it seems to make the most sense to just buy stuff and see what happens. Maybe nothing is wrong at all with the installed cabling and you've just gotten bad data due to an old machine and/or crappy USB adapter. Most other choices are economically foolish, unlikely, or just wrong.

Thanks! And I agree with your last statement. My inclination is to go down the route of purchasing the cams anyway and see what happens. Partly because over the past few days I've resigned myself to having new cables run in the worst case scenario. It won't cost me anything because I included running of CAT6 cable in the building work spec I've been quoted for, because I didn't know what the cable tests would show. The only downside is visible runs if conduit but I'm sure that can be minimised.

The other thing is, even if the cables are good, they are not outdoor rated so would they survive over time if I ensure any exposed sections are protected in conduit? Or is that another reason to just have new cables run...

 

[zerach]

Thanks! And I agree with your last statement. My inclination is to go down the route of purchasing the cams anyway and see what happens. Partly because over the past few days I've resigned myself to having new cables run in the worst case scenario. It won't cost me anything because I included running of CAT6 cable in the building work spec I've been quoted for, because I didn't know what the cable tests would show. The only downside is visible runs if conduit but I'm sure that can be minimised.

The other thing is, even if the cables are good, they are not outdoor rated so would they survive over time if I ensure any exposed sections are protected in conduit? Or is that another reason to just have new cables run...

If you use electrical conduit and make sure continuous condiut and the ends are protected from water/weather you should be covered, same goes for the cable inside. Are you worried about sun or weather?

I agree with what others have said laptop is likely to blame if not downstream equipment (routers/switches). I would be inclined to agree with the others and not worry about ifperf3 results, given you can‘t obtain reliable hardware. Ifperf3 is for performance related benchmarks, wireshark output would be better to find out why the packet drops and general network health.

Usually I would perform wireshark first for detailed information on network then use iperf3 for tuning after the network is healthy. That is a higher skilled more effort and more technical then your probably willing to do yourself. I am a network expert have been for a few decades.
more often than not a significant packet drop isn‘t going to be the crimped ends, just use a good pair tester. Its going to be cable length, atteniuation\electrical interference, or equipment.

check to make sure your cable length is within standards, do not run network cables in parallel with electrical cables, if it cannot be avoided cross at 90 degrees and minimize the crossing of main electrical trunks -keep away from anything giving off electrical interference you would be surprised at how much trouble a network cable over a doorbell or other bare transformer can be. And finally check equipment compatability upgrade firmware and keep drivers up to date. All good basic advice before going down the rabit hole of wireshark and iperf. Cable run replacement of a cat 5e or 6 cable is likely completely unnecessary unless its a cable break or 30 years old cat3/5 cable.

make sure if your going to do new runs that they use high quality shielded solid core cat6 cabling

hope your doing well in your project and good luck!

 

[freddyq]

If you use electrical conduit and make sure continuous condiut and the ends are protected from water/weather you should be covered, same goes for the cable inside. Are you worried about sun or weather?

I agree with what others have said laptop is likely to blame if not downstream equipment (routers/switches). I would be inclined to agree with the others and not worry about ifperf3 results, given you can‘t obtain reliable hardware. Ifperf3 is for performance related benchmarks, wireshark output would be better to find out why the packet drops and general network health.

Usually I would perform wireshark first for detailed information on network then use iperf3 for tuning after the network is healthy. That is a higher skilled more effort and more technical then your probably willing to do yourself. I am a network expert have been for a few decades.
more often than not a significant packet drop isn‘t going to be the crimped ends, just use a good pair tester. Its going to be cable length, atteniuation\electrical interference, or equipment.

check to make sure your cable length is within standards, do not run network cables in parallel with electrical cables, if it cannot be avoided cross at 90 degrees and minimize the crossing of main electrical trunks -keep away from anything giving off electrical interference you would be surprised at how much trouble a network cable over a doorbell or other bare transformer can be. And finally check equipment compatability upgrade firmware and keep drivers up to date. All good basic advice before going down the rabit hole of wireshark and iperf. Cable run replacement of a cat 5e or 6 cable is likely completely unnecessary unless its a cable break or 30 years old cat3/5 cable.

make sure if your going to do new runs that they use high quality shielded solid core cat6 cabling

hope your doing well in your project and good luck!

The cable has been rolled up and hung at the points outside the so concern would be how both sun and rain/cold temps have affected it..

I don't mind using wireshark if it could help me pinpoint the issue and potentially avoid running new cables. How does it work?

In terms of the causes you mention for packet loss I can fairly confidently knock out cable length as we're talking short runs here of 3-4 metres up to maybe 12 or 15 for the longest one. Electrical interference could be I guess and equipment seems most likely given I have to use a USB adapter on one end and a very old laptop on the other end. Only issue is I can't check the cable in between the ends because it was run through ceilings and floors when the house was done.

I've recently taken delivery of a couple of cameras from Andy and I've setup the camera using a machine terminated cat6 cable connected to my poe switch and all worked fine. So now I've done that, I'm going to connect the same camera to each of my cables and see what happens. This cable is high quality UTP solid core CAT6 so I do want to give it the best chance before I do reruns which can never be as clean as the cable I've already had run.



Sent from my SM-G950F using Tapatalk

 

[zerach]

The cable has been rolled up and hung at the points outside the so concern would be how both sun and rain/cold temps have affected it..

I don't mind using wireshark if it could help me pinpoint the issue and potentially avoid running new cables. How does it work?

In terms of the causes you mention for packet loss I can fairly confidently knock out cable length as we're talking short runs here of 3-4 metres up to maybe 12 or 15 for the longest one. Electrical interference could be I guess and equipment seems most likely given I have to use a USB adapter on one end and a very old laptop on the other end. Only issue is I can't check the cable in between the ends because it was run through ceilings and floors when the house was done.

I've recently taken delivery of a couple of cameras from Andy and I've setup the camera using a machine terminated cat6 cable connected to my poe switch and all worked fine. So now I've done that, I'm going to connect the same camera to each of my cables and see what happens. This cable is high quality UTP solid core CAT6 so I do want to give it the best chance before I do reruns which can never be as clean as the cable I've already had run.



Sent from my SM-G950F using Tapatalk

If you checked cables with a toner and all pairs are good. Thats all you need to do with cables.

I agree at this point hook up the cameras and check for throughput of the signal, if its expected throughput and and good picture no issue. If you still have an issue then start using wireshark. If you suspect packet drops being an issue wireshark is the way to narrow it down. However this is a skill thats not for the uninitiated. this involves education and experience, watch youtube there is tons of wireshark stuff itll give you the basics. H

I would much rather take my spare pile of network equipment and do replacement Tests old switches are sufficient enough to check picture quality issues etc. once you determine equipment replacement then go to store to get what you want.

theres basic things you can start with wire sniffing points before and after equipment and do simple packet count analysis, 112 packes on 1 end Of a switch 108 on the other, switch dropping a packet. Finding out why is a major task if its consumer equipment just replace it with comparable or better. if its high end equipment check logs and vendor sites for firmware updates etc.

if its more invovled, then its $$$ to troubleshoot network specialists aren’t cheap.

$80 switch replacement vs $150-$200 per hour rate. Common sense dictates replace equipment, upgrade firmware

If you need more help and the above doesnt pan out PM me and we will see what we can work out for testing.

 

[freddyq]

If you checked cables with a toner and all pairs are good. Thats all you need to do with cables.

I agree at this point hook up the cameras and check for throughput of the signal, if its expected throughput and and good picture no issue. If you still have an issue then start using wireshark. If you suspect packet drops being an issue wireshark is the way to narrow it down. However this is a skill thats not for the uninitiated. this involves education and experience, watch youtube there is tons of wireshark stuff itll give you the basics. H

I would much rather take my spare pile of network equipment and do replacement Tests old switches are sufficient enough to check picture quality issues etc. once you determine equipment replacement then go to store to get what you want.

theres basic things you can start with wire sniffing points before and after equipment and do simple packet count analysis, 112 packes on 1 end Of a switch 108 on the other, switch dropping a packet. Finding out why is a major task if its consumer equipment just replace it with comparable or better. if its high end equipment check logs and vendor sites for firmware updates etc.

if its more invovled, then its $$$ to troubleshoot network specialists aren’t cheap.

$80 switch replacement vs $150-$200 per hour rate. Common sense dictates replace equipment, upgrade firmware

If you need more help and the above doesnt pan out PM me and we will see what we can work out for testing.

Many thanks for this. Only question I have is, how do I check throughput of the signal once I have the camera connected to the cable? Is there a scientific way to do it or is it simply looking for quality issues in the footage being relayed?

The switch I'm using is actually not too old, I bought it in 2016 and it was this one. I'm confident the switch hardware is highly unlikely to blame with this packet loss situation and it's the same switch I used in the test I did when I connected up the camera with a machine-terminated cable.

 

[zerach]

Depends on program or the equipment. Usually I would do a video quality tune to constant bitrate then use the software to ensure the bps is correct and the video quality is good. My hardware has a basic network rate per camera monitor that makes this easy

It is likely that old laptop is to blame, if it was dropping packets without the switch. Are you planning on using some other hardware to monitor the camera feeds?

your average 100Mb switch will handle ~6-10 cameras at 4-6 mbps rates pretty easily, hardware has to keep up. GB switches are nice to have on shared or multiple networks for better packet handling theres always 2 primary reasons to go higher processing on switches bandwidth and the greatly missed packet counts. You can have low bandwidth and high packet counts and still saturate a switch due to high transaction rates.

Anything beyond that we are in to wireshark and the big mystery of network troubleshooting.

P.S. Netgear is a great choice for switches they are of nice quality and they always give you a good processing rate for small business to medium business switches. That one would be better than the average consumer switch and works well. I have 3 of their switches and never run in to a problem. That one has a 16Gbps backplane so it’ll have no problem handling all ports at full throughput. One of the reasons I like netgear, most middle of the road switches wont tell you the backplane throughput.

 

[freddyq]

Depends on program or the equipment. Usually I would do a video quality tune to constant bitrate then use the software to ensure the bps is correct and the video quality is good. My hardware has a basic network rate per camera monitor that makes this easy

It is likely that old laptop is to blame, if it was dropping packets without the switch. Are you planning on using some other hardware to monitor the camera feeds?

your average 100Mb switch will handle ~6-10 cameras at 4-6 mbps rates pretty easily, hardware has to keep up. GB switches are nice to have on shared or multiple networks for better packet handling theres always 2 primary reasons to go higher processing on switches bandwidth and the greatly missed packet counts. You can have low bandwidth and high packet counts and still saturate a switch due to high transaction rates.

Anything beyond that we are in to wireshark and the big mystery of network troubleshooting.

P.S. Netgear is a great choice for switches they are of nice quality and they always give you a good processing rate for small business to medium business switches. That one would be better than the average consumer switch and works well. I have 3 of their switches and never run in to a problem. That one has a 16Gbps backplane so it’ll have no problem handling all ports at full throughput. One of the reasons I like netgear, most middle of the road switches wont tell you the backplane throughput.

Interesting! So for now, because I've got some work going on at the house I thought it was a good opportunity to finally get cameras installed on these cables I've had hanging outside the house for so long. Once things settle down and finances allow, I'll be getting myself an efficient PC I can run BI on and that's how I'll be monitoring and recording the camera feeds. But I don't have the PC or BI at the moment however, I do need to ensure the cameras are good on these cables so that if not, I can run new cables while the work is still going on.

So is there a way I can check the throughput of the signal to the camera without BI or a PC?

And thanks for the info on my Netgear switch - good to get confirmation that hardware is up to the task.

 

[zerach]

Interesting! So for now, because I've got some work going on at the house I thought it was a good opportunity to finally get cameras installed on these cables I've had hanging outside the house for so long. Once things settle down and finances allow, I'll be getting myself an efficient PC I can run BI on and that's how I'll be monitoring and recording the camera feeds. But I don't have the PC or BI at the moment however, I do need to ensure the cameras are good on these cables so that if not, I can run new cables while the work is still going on.

So is there a way I can check the throughput of the signal to the camera without BI or a PC?

And thanks for the info on my Netgear switch - good to get confirmation that hardware is up to the task.

camera -> switch -> laptop with wireshark use the analyse-> bandwith\throughput feature. For 1 camera that laptop you mentioned will do.

heres a quick link on how to do it, scroll down To the monitoring section see the charts on the io monitor.

wiki.zeromq.org" data-pending="false">

Traffic Monitoring - zeromq

wiki.zeromq.org

read the info there, you’ll get a sense of how it gets set up.
required info:

- if address of camera once powered on - default ip is fine.
- port information
- configure laptop to connect to the camera network
- use a browser to connect to camera configure cbr (constant bit rate) on each of the streams Main/Sub I did 1040
- Go to the live tab and make sure you have the right channel you want to monitor main/sub
- load up wireshark and capture packets on laptop
- use wireshark Capture stats like screenshot below
- visually inspect your video stream make sure it looks good.
- achieve your happy place


Update: Best way I found to do this for basic checking is through statistics -> conversations using a capture filter of "host <ip camera ip> and port 80" <- with out the quotes


My settings were CBR 10240KB/sec Main and 1024K Sub as you can see the 10M bits /sec is the same metric (Red Box). If you watch the lower 1024K sub - you'll notice it's not exact and it'll be 1024 - 1200K/sec due to webpage overhead. But it's a thumbwag and not meant for precision accuracy. if you switch streams make sure you reset the capture. It'll show the conversations cleanly like below - obviously the most packets and Bytes is the video stream. If you let it run you'll end of clicking on things and quickly loading up 50 or more streams and it'll be a needle in a haystack for you.



Substream 1 CBR 1024K/sec



This will be the least CPU intensive/Network intensive way of doing it on that old laptop. If you just want to use a webbrowser on that network - you could also get an access point and tie in to the network wirelessly with a phone or other tablet device. But then you would need the direct stream URL to get in to it directly:

Connect to Dahua IP cameras

Note: Don't ever put in your real password in to this generator, just modify it after - I don't trust the site enough, it could record it.

 

[freddyq]

camera -> switch -> laptop with wireshark use the analyse-> bandwith\throughput feature. For 1 camera that laptop you mentioned will do.

heres a quick link on how to do it, scroll down To the monitoring section see the charts on the io monitor.

wiki.zeromq.org" data-pending="false">

Traffic Monitoring - zeromq

wiki.zeromq.org

read the info there, you’ll get a sense of how it gets set up.
required info:

- if address of camera once powered on - default ip is fine.
- port information
- configure laptop to connect to the camera network
- use a browser to connect to camera configure cbr (constant bit rate) on each of the streams Main/Sub I did 1040
- Go to the live tab and make sure you have the right channel you want to monitor main/sub
- load up wireshark and capture packets on laptop
- use wireshark Capture stats like screenshot below
- visually inspect your video stream make sure it looks good.
- achieve your happy place


Update: Best way I found to do this for basic checking is through statistics -> conversations using a capture filter of "host and port 80" B]


My settings were CBR 10240KB/sec Main and 1024K Sub as you can see the 10M bits /sec is the same metric (Red Box). If you watch the lower 1024K sub - you'll notice it's not exact and it'll be 1024 - 1200K/sec due to webpage overhead. But it's a thumbwag and not meant for precision accuracy. if you switch streams make sure you reset the capture. It'll show the conversations cleanly like below - obviously the most packets and Bytes is the video stream. If you let it run you'll end of clicking on things and quickly loading up 50 or more streams and it'll be a needle in a haystack for you.

View attachment 51975

Substream 1 CBR 1024K/sec

View attachment 51976

This will be the least CPU intensive/Network intensive way of doing it on that old laptop. If you just want to use a webbrowser on that network - you could also get an access point and tie in to the network wirelessly with a phone or other tablet device. But then you would need the direct stream URL to get in to it directly:

Connect to Dahua IP cameras

Note: Don't ever put in your real password in to this generator, just modify it after - I don't trust the site enough, it could record it.

So I hooked up the camera to two of the cables, one after another to test focal lengths and the video stream looked absolutely fine on both cables. No lagging or video quality issues at all. So that's a good result.

Must admit the wireshark stuff isn't making much sense to me but I think I need to spend more time on it and I should really do it to confirm once and for all if these cables are good to use for the cameras.

Anyway, the dumb non scientific test has at least yielded very positive results!

Sent from my SM-G950F using Tapatalk

 

[freddyq]

Slightly off-topic and newbie question but related to my setup, is there any difference in performance from the perspective of running cameras in the following two scenarios.

I have CAT6 cable running to a location where my extension is currently being built. The other end of this CAT6 cable is connected to a switch which in turn is connected to my router. So at the new extension location:

1. Terminate CAT6 cable at an ethernet wall socket then plug a PoE switch into that wall socket and run my cams from that switch
2. Just plug the CAT6 cable straight into a PoE switch (without terminating at a wall socket) and then run my cams from that switch

Obviously, scenario 1 is neater but I'm just not sure if creating a break in the run by introducing the wall socket makes any difference or not. If it does, I have enough length of cable and opportunity to hide under floors etc. to run it straight to the point in my new room where it could plug into the PoE switch, if that's the better option.

 

[mikeynags]

Either option is good. No, you won't create any issues with the break introduced by the wall socket as you mention above. No such thing.

 

[freddyq]

Either option is good. No, you won't create any issues with the break introduced by the wall socket as you mention above. No such thing.

Thanks for confirming!

Sent from my SM-G950F using Tapatalk

 

[freddyq]

Just an update on the original issue of this thread - I still haven't had time to get to grips with Wireshark so I can properly test those cables but what I have done is connected a camera to each one and the good news is that the camera was detected on my network and I was able to view a perfectly good live stream on all the cables. So it seems the cables are absolutely fine which is great news.

Even if I don't get time to do the Wireshark testing I can definitely be more confident of using the existing cables for my camera network.