03-21-2012 10:34 AM - edited 03-04-2019 03:44 PM
Hello all,
Wanting to get some advice/help! I have just received Time Warner Cable and get 16Mbps down from them directly through their modem (motorola DOCSIS 2.0 modem- SB5101) to my PC (using speedtest.net). When I then connect TWC's modem through my 2801 router on fa0/0, the download rate I get is considerably slower (on the order of 3-4Mbps via same testing method). What am I missing?
Here is how the passthrough interface is setup:
interface FastEthernet0/0
bandwidth receive inherit
ip address dhcp
ip nat outside
ip virtual-reassembly in
duplex auto
speed auto
I have attempted removing bandwidth command - nothing changed with it not there. I have also hard coded the speed and duplex to 100/full to no avail....
Other possible relevant configs:
ip nat inside source list 1 interface FastEthernet0/0 overload
ip route 0.0.0.0 0.0.0.0 dhcp
access-list 1 permit 192.168.1.0 0.0.0.255
Basically this things is wide open... for testing but the download speed differential is perplexing to me.
Is it this? Or some thing else? Thanks in advance for your help!
tagman25
Solved! Go to Solution.
03-22-2012 11:53 AM
03-21-2012 10:55 AM
Disclaimer
The Author of this posting offers the information contained within this posting without consideration and with the reader's understanding that there's no implied or expressed suitability or fitness for any purpose. Information provided is for informational purposes only and should not be construed as rendering professional advice of any kind. Usage of this posting's information is solely at reader's own risk.
Liability Disclaimer
In no event shall Author be liable for any damages whatsoever (including, without limitation, damages for loss of use, data or profit) arising out of the use or inability to use the posting's information even if Author has been advised of the possibility of such damage.
Posting
What's the router CPU usage look like during these speedtests?
03-22-2012 06:40 AM
The cpu process does not go above 4% during the speed tests. Confirmed that last night. ANy other thoughts. I cannot imagine that NAT would reduce the speed that much. The reasoning behind that is because I just got rid of 5Mbps download speeds with another company, and the speedtests showed around the 3-4Mbps speed when I had that as well.
ANY other thoughts, advise, help would be appreciated. Thank you.
03-22-2012 07:55 AM
Here is a Cisco performance guide that covers the 2800 series
http://www.cisco.com/web/partners/downloads/765/tools/quickreference/routerperformance.pdf
As you can see, process switching is slow.
03-23-2012 03:02 AM
Disclaimer
The Author of this posting offers the information contained within this posting without consideration and with the reader's understanding that there's no implied or expressed suitability or fitness for any purpose. Information provided is for informational purposes only and should not be construed as rendering professional advice of any kind. Usage of this posting's information is solely at reader's own risk.
Liability Disclaimer
In no event shall Author be liable for any damages whatsoever (including, without limitation, damages for loss of use, data or profit) arising out of the use or inability to use the posting's information even if Author has been advised of the possibility of such damage.
Posting
The only other thing that comes to mind, is speedtests often only run for a few seconds and the router's CPU might not accurately reflect the short term load. However, assuming that's not the case, the only other thing that comes to mind, depending on the test, if the router is dropping any packets might degrade test performance.
I've never intentionally stressed tested a 2801 to its maximum but I have the next larger model in the series, the 2811. On this model, I was able to push about 40 Mbps unidirectional UDP traffic. If performance ratios are as I would expect, the 2801 should have enough performance to deal with 20 Mbps (aggregate up/down?) aggregate on your cable link.
03-21-2012 11:21 AM
I will have a result on the cpu usage this evening, however if I remember correctly - it was never above 4%. Confirmation coming this evening.
03-21-2012 12:35 PM
With NAT, you are process-switching the packets. Regardless of the router, thats the slowest method of all
here is a link to use for some tuning at the workstation:
http://cisconet.com/traffic-analysis/throughput/104-tcp-throughput-calculation-formula.html
03-22-2012 11:53 AM
Hi
virtual reasembly can be a problem for performance.
HTH Patrick
03-23-2012 12:10 PM
I took out the 'ip virtual reassembly in' in the interface and started receiving WAY faster download rates. Thank you!
03-23-2012 01:14 PM
Disclaimer
The Author of this posting offers the information contained within this posting without consideration and with the reader's understanding that there's no implied or expressed suitability or fitness for any purpose. Information provided is for informational purposes only and should not be construed as rendering professional advice of any kind. Usage of this posting's information is solely at reader's own risk.
Liability Disclaimer
In no event shall Author be liable for any damages whatsoever (including, without limitation, damages for loss of use, data or profit) arising out of the use or inability to use the posting's information even if Author has been advised of the possibility of such damage.
Posting
The feature is documented as: (from http://www.cisco.com/en/US/docs/ios/12_3t/12_3t8/feature/guide/gt_vfrag.html)
Performance Impact
VFR will cause a performance impact on the basis of functions such as packet copying, fragment validation, and fragment reorder. This performance impact will vary depending on the number of concurrent IP datagram that are being reassembled.
VFR is designed to work with any feature that requires fragment reassembly (such as Cisco IOS Firewall and NAT). Currently, NAT enables and disables VFR internally; that is, when NAT is enabled on an interface, VFR is automatically enabled on that interface.
However, surprised this doesn't show a difference for CPU usage, especially as it's noticeable impact to performance, as you've described. Also surprising if you were NOT dealing with fragments, although the above does mention packet copying. Glad Patrick noted this.
03-24-2012 03:34 AM
Thanks no Problem.
I noticed it some time ago on a 3640:-) router can do 80 - 100 Mbit routing over two FastEthers. With VirualReaembly it drops to 2 Mbit/s.
VPN encryption on them was also crappy..
Alway be carefull with Features.
2801 should do 40 Mbit/s plus. Bad Traffic Mix (64k Packts and Services drops performance)
Discover and save your favorite ideas. Come back to expert answers, step-by-step guides, recent topics, and more.
New here? Get started with these tips. How to use Community New member guide