Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[displaced]

I can really recommend running an iperf3 (https://iperf.fr/iperf-download.php) server or two on your network.

iperf3 can run a range of IP transfer tests across your network.  I've used it on FreeBSD, MacOS and Win10, so it'll run on whatever you've got!

I used it to test my gigabit cabling soon after installation to ensure it was behaving correctly under load.  Another good use is to check that switches and routers are capable of saturating all of their ports.  For example, a cheap 5-port gigabit switch might not necessarily be capable of pushing 5Gbit across all ports simultaneously. 

So, perhaps install an iperf3 server on a couple of hosts, then run the iperf3 client on some of your other hosts in various combinations to thoroughly exercise your networking equipment.

It's quite satisfying to make the LEDs on a few gigabit switches scream

[displaced]

One question :  Does all data get routed through the main router on a network like mine?

I always assumed that all traffic goes via the router which is why I thought it would be better to have the server on the main router (not the switch) to avoid having the link between the router and the switch (LAN 5) having to cope with double the data rate for each stream i.e. up to the router and then back to the client each time ...... is this wrong?

Understandable that you'd think that, but it is indeed wrong

Routers route between networks.  Switches direct traffic to the destination host within the same network.  If you have a 5-port switch with your router and two hosts (A, B) connected, the switch will see traffic from Host A destined for Host B and direct the incoming traffic to the port to which Host B is connected.  It'll never travel back to the router first.

The ethernet ports on the router itself behave in the same way.  The 'router' part of the router (pardon the clumsy phrasing!) only gets involved if a packet is destined for an address outside of your home network.  Such packets are sent to the router (your 'default gateway' for packets not destined for your home network) which passes the data to your ISP for onward travel.  The router then needs to keep track of those outbound connections so that it can route replies back to the internal host which initiated the request.

Hope that helps!

[edit]: just to add:  If it helps, think of it like this:  If devices are connected to the same gigabit switch, and the switch is decent, you'll have 1Gb of bandwidth between all those devices at all times, with no contention.  When a device connected to switch A talks to a device on switch B, you have a maximum of 1Gb but you'll be contending for that 1Gb with other devices talking to eachother between those switches.

[d2d4j]

Hi

If the DNS is controlled by the router, then the router is involved at first, but this may not be true if there is another device used for DNS inside the network

Many thanks

John

[displaced]

Hi

If the DNS is controlled by the router, then the router is involved at first, but this may not be true if there is another device used for DNS inside the network

Very true, though DNS requests are teeny-tiny, and most clients will cache those quite aggressively once done.

Oh -- and Chunkers -- I saw an 'o7' salute back there ... are you an Elite: Dangerous player by any chance?

[Chunkers]

Understandable that you'd think that, but it is indeed wrong

Wow, I am much clearer now thanks so much to you guys for the explanation.  I am now feeling a bit bad about cheaping-out on my switches, I guess I have always just seen them as "extra holes" to plug into and that the router did all the work but now I know they do "stuff" I kinda wished I had gone for quality not cheapness, lol. 

I can really recommend running an iperf3 (https://iperf.fr/iperf-download.php) server or two on your network.

I am going to give this a go, definitely, but it will be after I get back from work as I will be away for 4 weeks from Monday.

I am embarrassed at my own ignorance and am going to try and improve my knowledge of how networks, er ..... work

I saw an 'o7' salute back there ... are you an Elite: Dangerous player by any chance?

I do have a copy, in fact I bought the game before it was released, and I think I would like it because I am old enough to have been obsessed with the original - I even went out and bought a HOTAS joystick to play it with but haven't got round to it yet.

In fact I am an Eve Online player, have been playing since 2007, a game which has consumed vast numbers of hours of my life.  Its also a spaceship game, but from what I have seen, more strategic and with a lot more "metagaming".  Its pretty much all I have played for the last 9 years, I occasionally flirt with FPS games, however, currently playing SWBF which is great fun and just bought BF1.

Its a great time to try Eve, actually, CCP has just released a patch allowing "Alpha state" clones to play for free - the learning curve is extreme but don't let it put you off - if anyone wants to give it a try then let me know.

We should have a "what games do you play" thread (unless there is one already)

o7

Chunks

[displaced]

Wow, I am much clearer now thanks so much to you guys for the explanation.  I am now feeling a bit bad about cheaping-out on my switches, I guess I have always just seen them as "extra holes" to plug into and that the router did all the work but now I know they do "stuff" I kinda wished I had gone for quality not cheapness, lol. 

No problem!  And not to worry -- it's probably just as bad to pay way over the odds for some enterprise-grade kit when all you need is something just a bit better than bog-standard.  Now you know how the data's physically moving around, it might become clear how you want your network structured. 

Even if you rip it all out and end up putting it back exactly how it was, it's all fun, eh?  (...and that's why I get in trouble with the Mrs. )

Personally, I've had really good experiences with the Netgear GS105E (5-port) and GS108E (8-port) switches.  They're both capable of pushing full bandwidth across all ports simultaneously and have some handy features that are a bit above consumer level without costing 'pro' prices.

I am going to give this a go, definitely, but it will be after I get back from work as I will be away for 4 weeks from Monday.

Cool -- well if you need any help running iperf3 give me a shout.  It really is a great way to pressure your LAN in a controlled way to see how it performs.

I am embarrassed at my own ignorance and am going to try and improve my knowledge of how networks, er ..... work

Hehe -- don't worry about that at all. More fun to learn as you go along

I do have a copy, in fact I bought the game before it was released, and I think I would like it because I am old enough to have been obsessed with the original - I even went out and bought a HOTAS joystick to play it with but haven't got round to it yet.

Same here on the age thing.  Elite on the Acorn, then Frontier on the Atari ST.  I too got a HOTAS and have lost far too much time to E:D!  Never quite got into Eve, but do enjoy reading the stories that come out of the community!

Good luck!

Chris

[Chunkers]

I can really recommend running an iperf3 (https://iperf.fr/iperf-download.php) server or two on your network.

iperf3 can run a range of IP transfer tests across your network.  I've used it on FreeBSD, MacOS and Win10, so it'll run on whatever you've got!

I used it to test my gigabit cabling soon after installation to ensure it was behaving correctly under load.  Another good use is to check that switches and routers are capable of saturating all of their ports.  For example, a cheap 5-port gigabit switch might not necessarily be capable of pushing 5Gbit across all ports simultaneously. 

So, perhaps install an iperf3 server on a couple of hosts, then run the iperf3 client on some of your other hosts in various combinations to thoroughly exercise your networking equipment.

It's quite satisfying to make the LEDs on a few gigabit switches scream

Just an addendum to this moderately old thread I installed iperf3 and tested my server and a couple of wired clients on my network and things seem to be OK.

I really like iperf as a tool, it works great and is super easy to use, thanks for the recommendation @displaced.

It looks to me like there is not much wrong between my server and the clients if I am reading it right, I have read the theoretical maximum for a Gb network is 125 Mbit/s so I am 80% there :

C:\iperf>iperf3 -s
-----------------------------------------------------------
Server listening on 5201
-----------------------------------------------------------
Accepted connection from 192.168.1.10, port 56106
[  5] local 192.168.1.100 port 5201 connected to 192.168.1.10 port 56107
[ ID] Interval           Transfer     Bandwidth
[  5]   0.00-1.00   sec  10.9 MBytes  91.2 Mbits/sec
[  5]   1.00-2.00   sec  11.3 MBytes  94.9 Mbits/sec
[  5]   2.00-3.00   sec  11.3 MBytes  94.9 Mbits/sec
[  5]   3.00-4.00   sec  11.3 MBytes  94.9 Mbits/sec
[  5]   4.00-5.00   sec  11.3 MBytes  94.9 Mbits/sec
[  5]   5.00-6.00   sec  11.3 MBytes  95.0 Mbits/sec
[  5]   6.00-7.00   sec  11.3 MBytes  94.9 Mbits/sec
[  5]   7.00-8.00   sec  11.3 MBytes  94.9 Mbits/sec
[  5]   8.00-9.00   sec  11.3 MBytes  94.9 Mbits/sec
[  5]   9.00-10.00  sec  11.3 MBytes  94.9 Mbits/sec
[  5]  10.00-10.06  sec   637 KBytes  93.8 Mbits/sec
- - - - - - - - - - - - - - - - - - - - - - - - -
[ ID] Interval           Transfer     Bandwidth
[  5]   0.00-10.06  sec  0.00 Bytes  0.00 bits/sec                  sender
[  5]   0.00-10.06  sec   113 MBytes  94.5 Mbits/sec                  receiver
-----------------------------------------------------------
Server listening on 5201
-----------------------------------------------------------
Accepted connection from 192.168.1.11, port 63616
[  5] local 192.168.1.100 port 5201 connected to 192.168.1.11 port 63617
[ ID] Interval           Transfer     Bandwidth
[  5]   0.00-1.00   sec  10.7 MBytes  89.5 Mbits/sec
[  5]   1.00-2.00   sec  11.3 MBytes  94.9 Mbits/sec
[  5]   2.00-3.00   sec  11.3 MBytes  94.5 Mbits/sec
[  5]   3.00-4.00   sec  11.3 MBytes  94.5 Mbits/sec
[  5]   4.00-5.00   sec  11.3 MBytes  94.9 Mbits/sec
[  5]   5.00-6.00   sec  11.3 MBytes  95.0 Mbits/sec
[  5]   6.00-7.00   sec  11.3 MBytes  94.9 Mbits/sec
[  5]   7.00-8.00   sec  11.3 MBytes  95.0 Mbits/sec
[  5]   8.00-9.00   sec  11.3 MBytes  94.9 Mbits/sec
[  5]   9.00-10.00  sec  11.3 MBytes  94.5 Mbits/sec
[  5]  10.00-10.07  sec   858 KBytes  94.1 Mbits/sec
- - - - - - - - - - - - - - - - - - - - - - - - -
[ ID] Interval           Transfer     Bandwidth
[  5]   0.00-10.07  sec  0.00 Bytes  0.00 bits/sec                  sender
[  5]   0.00-10.07  sec   113 MBytes  94.3 Mbits/sec                  receiver
-----------------------------------------------------------
Server listening on 5201
-----------------------------------------------------------

Chunks

[displaced]

Glad you've found iperf3 useful!  It's great for seeing how many bits you can shove around a network without worrying about higher-level protocols.

However, your stats are a bit off.  Gigabit should be 100 Mbytes/sec, not Mbits - it looks like you've got some congestion somewhere or a hardware problem (faulty Gbit ethernet cable, under-spec router, etc).

Here's what it looks like on my Gigabit LAN, going from my office PC, via two Netgear switches (a GS108E and a GS105E) to my server:

Code: [Select]

Connecting to host trillian.home, port 5201
[  4] local 192.168.50.13 port 58119 connected to 192.168.50.50 port 5201
[ ID] Interval           Transfer     Bandwidth
[  4]   0.00-1.00   sec   109 MBytes   911 Mbits/sec
[  4]   1.00-2.00   sec   102 MBytes   859 Mbits/sec
[  4]   2.00-3.00   sec  99.0 MBytes   831 Mbits/sec
[  4]   3.00-4.00   sec   102 MBytes   860 Mbits/sec
[  4]   4.00-5.00   sec   110 MBytes   925 Mbits/sec
[  4]   5.00-6.00   sec   108 MBytes   906 Mbits/sec
[  4]   6.00-7.00   sec   109 MBytes   913 Mbits/sec
[  4]   7.00-8.00   sec   110 MBytes   926 Mbits/sec
[  4]   8.00-9.00   sec   108 MBytes   910 Mbits/sec
[  4]   9.00-10.00  sec   108 MBytes   910 Mbits/sec
- - - - - - - - - - - - - - - - - - - - - - - - -
[ ID] Interval           Transfer     Bandwidth
[  4]   0.00-10.00  sec  1.04 GBytes   895 Mbits/sec                  sender
[  4]   0.00-10.00  sec  1.04 GBytes   895 Mbits/sec                  receiver

iperf Done.

Your numbers are, I think, pretty much spot-on for a 100Mbit network -- so perhaps it's worth double-checking all network adapter stats/settings and the LEDs on your routers to check that all links are indicating 1 Gigabit. 

Also, some of your network gear (either the adapters in your machines or the switches themselves) may provide cable testing functionality.  Modern equipment will fall back to slower speeds if the cable is out-of-spec, rather than not work at all.

Good luck!

[d2d4j]

Hi

Yes,I concur with @displaced, in that your stats look like 100 network.  What is the path taken between those 2 pc.  I am thinking same, that 1 device or more maybe downgrading to 100 from a 1000, which as displaced has posted, could be device or cable, it could even be on the switch, if set to 100, and not 1000

If your interested, here is a live test from our speedtest server, but to test to this level on internet, you would need 1GB connection to internet

Many thanks

John

« SpeedGuide.net Speed Test Results »
1398101 kbps down (~1398.1 Mbps, 170667 KB/s)↓
88 kbps up (~0.09 Mbps, 11 KB/s)↑
10240 KB downloaded in 0.06 seconds
4 KB uploaded in 0.371 seconds
Tested on: 2016.12.12 03:41 EST
Tested from: helweb.co.uk
Test Link: http://www.speedguide.net/speedtest/results.php?test=4619345
Latency: 133ms
Provider: t-ipconnect.de
Location: DE

[Chunkers]

OK, I am glad I posted these results and thanks for the replies, I have all green lights on my router indicating Gb connections so I have some homework to do .....

hehe

Chunks

[Chunkers]

WOW, I THINK WE MIGHT HAVE CRACKED IT BOYZ

I couldn't think of any other reason so I replaced the virtually new CAT6 network cable to my server with a horrible old CAT 5E one and hey presto!

Windows PowerShell
Copyright (C) 2016 Microsoft Corporation. All rights reserved.

PS C:\Users\Jimjams> wmic NIC where "NetEnabled='true'" get "Name","Speed"
Name                                Speed
Realtek PCIe GBE Family Controller  1000000000

PS C:\Users\Jimjams>

And also iperf look more like what you guys said :

C:\iperf>iperf3 -c 192.168.1.100
Connecting to host 192.168.1.100, port 5201
[  4] local 192.168.1.10 port 51445 connected to 192.168.1.100 port 5201
[ ID] Interval           Transfer     Bandwidth
[  4]   0.00-1.00   sec   107 MBytes   901 Mbits/sec
[  4]   1.00-2.00   sec   112 MBytes   940 Mbits/sec
[  4]   2.00-3.00   sec   112 MBytes   938 Mbits/sec
[  4]   3.00-4.00   sec   112 MBytes   938 Mbits/sec
[  4]   4.00-5.00   sec   111 MBytes   934 Mbits/sec
[  4]   5.00-6.00   sec   112 MBytes   936 Mbits/sec
[  4]   6.00-7.00   sec   112 MBytes   938 Mbits/sec
[  4]   7.00-8.00   sec   111 MBytes   931 Mbits/sec
[  4]   8.00-9.00   sec   111 MBytes   932 Mbits/sec
[  4]   9.00-10.00  sec   112 MBytes   935 Mbits/sec
- - - - - - - - - - - - - - - - - - - - - - - - -
[ ID] Interval           Transfer     Bandwidth
[  4]   0.00-10.00  sec  1.09 GBytes   932 Mbits/sec                  sender
[  4]   0.00-10.00  sec  1.09 GBytes   932 Mbits/sec                  receiver

iperf Done.

C:\iperf>

If this solves my TV streaming issues I may have to come round your house(s) and french kiss you, brace yourselves

Chunks

[d2d4j]

Hi chunkers

Haha a beer is more appreciated thank you.

Glad you resolved it

Interestingly, are you sure your cat6 is not CCA (copper coated ali)... might explain it and there's a lot of of CCA advertised, even as solid copper

Also, may be well advised to check all pc and replace as needed.

Horribly old cat5e maybe full copper

Many thanks

John

[displaced]

Hi Chunkers!

Glad you got it sorted.  I've had a couple of out-of-spec ethernet cables over the years.  Sometimes it was CCA as John notes, others it was a case of the twisted pairs being barely twisted at all.  I once bought a 50 metre reel of alleged Cat6 cable whose wires were entirely straight - no twists at all.  They even had the cheek to stamp 'CAT6 VALIDATED' on the outside insulation.  Gits.

Cat5e is perfectly valid for gigabit speeds.  The 'e' makes all the difference.  I've only run 6e around my house for a bit of future-proofing, but only for the fixed runs.  Otherwise, I've got plenty of 5e patch leads between my machines and switches.

Anyway -- do let us know if this has fixed your buffering problems!  A 100Mbit link is easily saturated by high-bitrate files, especially when adding in some protocol overhead.  Now you're running about 10 times faster, you should be fine