[SOLVED] Tests Related to Recent Increase in Bandwidth Loss Through Router

[daddydave]

Finally got a chance to do some wi-fi testing because I seem to be losing a lot of bandwidth these days unless I skip the router altogether. This seems to be the case regardless of whether I go through wi-fi or an ethernet connection through the router.

https://docs.google....n6zKrc/edit?hl=en_US

[mwb1100]

What's the router model?  Do you have an opportunity to test with another router, like by borrowing one from home or something?  Has the router had it's firmware updated recently (if so, maybe that's the cause; if not, maybe an update might help)?

[40hz]

OK. You've piqued my curiosity...

1) What are you using for a router?

2) What are the port speeds on each device? Is the router only 100Mbs (on either its WAN or LAN side) whereas the cable modem has 1Gb speed on its RJ45 jack?

What's puzzling is that you said you were seeing very high throughput previously with the router in circuit. So the problem is obviously related to the router being in the loop. Sounds like some bandwidth throttling or QoS feature has been woken up on the router. Have you recently upgraded the firmware in the router? If so, there's a chance the update has  switched something ON which used to be OFF by default.

Possibly a misconfigured MTU or other setting on the router...hmm...

Question: does  the cable modem have it's own switch built in or does it just provide a single IP address on the LAN side? If it provides DHCP services, you'd be better off just plugging a switch into it and forgoing  the router altogether.

When plugged straight into the cable modem could you run an IPCONFIG /ALL command and see if your NIC has a non-routable IP address? Also if the subnet is anything other than a /24 (i.e. 255.255.255.0)?

Long shot: some device plugged into your network (or one of the ports on the router itself) has gone south and is introducing 'chatter' or other noise which is forcing enough retransmits that it's dragging the entire network down. Since the performance drop is there whenever the router (and nothing else) is plugged in, it's a very real possibility.
 

-----------------

@daddydave - OK...it's 1:00AM and the server I was working on is back online (RAID rebuild!) so I'll be packing up and heading home. Let's regroup on this later this weekend.

[Stoic Joker]

What's puzzling is that you said you were seeing very high throughput previously with the router in circuit. So the problem is obviously related to the router being in the loop. Sounds like some bandwidth throttling or QoS feature has been woken up on the router. Have you recently upgraded the firmware in the router? If so, there's a chance the update has  switched something ON which used to be OFF by default.

-40hz (October 22, 2011, 12:01 AM)

Sounds about right to me ... The SPI firewalls on consumer grade hardware also have a bad habit of screwing the pooch.

--------------------------------------------------------

Generally for this type of problem I use a special "Line Stressing" utility (packet flooder - that I'm afraid to release)  to target different points in the device chain to see who collapses first.

--------------------------------------------------------

It might not hurt to just reset & reconfigure the router in case a power blip (or other brand of hiccup) has corrupted it's current configuration.

[daddydave]

Short answer:
It's a Linksys Netgear WNR845T WNR854T (thanks Stoic) , and as far as I know, it doesn't have any QoS settings.
There hasn't been a firmware upgrade from Linksys (or anyone, as best as I can tell) since 2008.
I'll check and see if there is an MTU setting.
Trying a reset to factory default settings is probably the last stop before buying a new router, and I'll try to get on with "guest mode" while I am at it.
I seem to have ruled out interference as a major factor in a drop in download speed from 20 Mbps to 6 Mbps unless the interference is happening inside the router itself.

[Stoic Joker]

O_o Um... Any chance that's actually a Netgear WNR854T?? It does have an SPI firewall according to the manual.

[daddydave]

I disabled the SPI firewall and unscientifically (because I haven't dismantled everything again)  it seems to have zero effect on tested download speed. One IP, one ethernet port on the cable modem. Factory reset, here I come.

[daddydave]

That was the most painless factory reset I'd ever done. I was able to get to the setup wirelessly afterwards. And no effect on the problem at hand.

[Stoic Joker]

Check the ping time both to and through the router. If there is a significant rise in the ping time (which there shouldn't be) then the router becomes an official prime suspect.

I've never worked with that specific model, but when googling the number I did see a lot of unhappy people screaming about a lack of performance.

What is it with the new stuff these days ... I'm running 30Mbps both directions through an antique (~8yr old) Linksys WRT54G that I'll probably never replace (because I'm afraid to) unless it dies.

[daddydave]

You mean like this?

C:\Users\daddy>ping 192.168.1.1   <--- the router

Pinging 192.168.1.1 with 32 bytes of data:
Reply from 192.168.1.1: bytes=32 time=5ms TTL=64
Reply from 192.168.1.1: bytes=32 time=4ms TTL=64
Reply from 192.168.1.1: bytes=32 time=4ms TTL=64
Reply from 192.168.1.1: bytes=32 time=4ms TTL=64

Ping statistics for 192.168.1.1:
    Packets: Sent = 4, Received = 4, Lost = 0 (0% loss),
Approximate round trip times in milli-seconds:
    Minimum = 4ms, Maximum = 5ms, Average = 4ms

C:\Users\daddy>ping 192.168.1.3 <--- sister in law's blackberry

Pinging 192.168.1.3 with 32 bytes of data:
Reply from 192.168.1.3: bytes=32 time=365ms TTL=128
Reply from 192.168.1.3: bytes=32 time=192ms TTL=128
Reply from 192.168.1.3: bytes=32 time=412ms TTL=128
Reply from 192.168.1.3: bytes=32 time=127ms TTL=128

Ping statistics for 192.168.1.3:
    Packets: Sent = 4, Received = 4, Lost = 0 (0% loss),
Approximate round trip times in milli-seconds:
    Minimum = 127ms, Maximum = 412ms, Average = 274ms

C:\Users\daddy>ping 192.168.1.4  <-- the Roku, I think

Pinging 192.168.1.4 with 32 bytes of data:
Reply from 192.168.1.4: bytes=32 time=3ms TTL=64
Reply from 192.168.1.4: bytes=32 time=4ms TTL=64
Reply from 192.168.1.4: bytes=32 time=3ms TTL=64
Reply from 192.168.1.4: bytes=32 time=3ms TTL=64

Ping statistics for 192.168.1.4:
    Packets: Sent = 4, Received = 4, Lost = 0 (0% loss),
Approximate round trip times in milli-seconds:
    Minimum = 3ms, Maximum = 4ms, Average = 3ms

[daddydave]

Don't really know if this means anything,  (Blackberry was kicked off during my original testing)

C:\Users\daddy>pathping 192.168.1.1

Tracing route to 192.168.1.1 over a maximum of 30 hops

  0  daddy-PC [192.168.1.8]
  1  192.168.1.1

Computing statistics for 25 seconds...
            Source to Here   This Node/Link
Hop  RTT    Lost/Sent = Pct  Lost/Sent = Pct  Address
  0                                           daddy-PC [192.168.1.8]
                                0/ 100 =  0%   |
  1    4ms     0/ 100 =  0%     0/ 100 =  0%  192.168.1.1

Trace complete.

C:\Users\daddy>pathping 192.168.1.3

Tracing route to 192.168.1.3 over a maximum of 30 hops

  0  daddy-PC [192.168.1.8]
  1  192.168.1.3

Computing statistics for 25 seconds...
            Source to Here   This Node/Link
Hop  RTT    Lost/Sent = Pct  Lost/Sent = Pct  Address
  0                                           daddy-PC [192.168.1.8]
                                0/ 100 =  0%   |
  1  116ms     0/ 100 =  0%     0/ 100 =  0%  192.168.1.3

Trace complete.

C:\Users\daddy>pathping 192.168.1.4

Tracing route to 192.168.1.4 over a maximum of 30 hops

  0  daddy-PC [192.168.1.8]
  1  192.168.1.4

Computing statistics for 25 seconds...
            Source to Here   This Node/Link
Hop  RTT    Lost/Sent = Pct  Lost/Sent = Pct  Address
  0                                           daddy-PC [192.168.1.8]
                                0/ 100 =  0%   |
  1    5ms     0/ 100 =  0%     0/ 100 =  0%  192.168.1.4

Trace complete.

[Stoic Joker]

Sort of. But I was more looking for the ping time directly to the router's WAN interface IP, from the LAN side. That way you rule out any other misc issues with the route to another target device, and focus strictly on the time differential for getting a packet through the router itself.

I usually prefer to do a ping flood (steady stream of large unfragmented packets), but the Windows ping utility isn't capable of that type of shenanigan. However you can bump the packet size a bit by using:

C:\Users\Stoic Joker>ping 192.168.0.1 -f -l 1024

Pinging 192.168.0.1 with 1024 bytes of data:
Reply from 192.168.0.1: bytes=1024 time=1ms TTL=64
Reply from 192.168.0.1: bytes=1024 time=2ms TTL=64
Reply from 192.168.0.1: bytes=1024 time=2ms TTL=64
Reply from 192.168.0.1: bytes=1024 time=1ms TTL=64

Ping statistics for 192.168.0.1:
    Packets: Sent = 4, Received = 4, Lost = 0 (0% loss),
Approximate round trip times in milli-seconds:
    Minimum = 1ms, Maximum = 2ms, Average = 1ms

C:\Users\Stoic Joker>

That should be enough to make the weak link wince a bit.

[daddydave]

C:\Users\daddy>ping 192.168.1.1 -f -l 1024

Pinging 192.168.1.1 with 1024 bytes of data:
Reply from 192.168.1.1: bytes=1024 time=29ms TTL=64
Reply from 192.168.1.1: bytes=1024 time=18ms TTL=64
Reply from 192.168.1.1: bytes=1024 time=7ms TTL=64
Reply from 192.168.1.1: bytes=1024 time=13ms TTL=64

Ping statistics for 192.168.1.1:
    Packets: Sent = 4, Received = 4, Lost = 0 (0% loss),
Approximate round trip times in milli-seconds:
    Minimum = 7ms, Maximum = 29ms, Average = 16ms

[daddydave]

Well, I haven't put security on it yet, but I just put a new router on it, and it is testing at >25 Mbps download speed.

UPDATE
Now I have put WPA2-Personal on it and I am getting the same.

 

Thanks to everyone for your help.