Strange Network Issue (Internet Cisco 3800 and ASA 515)
Guys,
I have an issue I can't figure out. We have a DS3 connection. Fiber to a MUX in the Data Center then to 3800 series router then to Cisco ASA.
We can get around 41 Mbps download but only 1 to 2 upload at the max and sometimes less than that. This translates to our customers having issues when getting large files from us via FTP or Video.
To narrow it down I've done the following.
Placed a switch between my firewall and my outside internet connection effectively bypassing my firewall and network. I gave myself and outside IP and ran the speed test. The test show the same results of a normal download and very slow upload of not more than 2 Mbps.
I've checked my Internet router for errors and there are virtually no dropped packets, no runts, no overruns, cpu utilization is 10% and memory is 40%. I've also looked at the buffers and the small buffers had maybe 300 misses but that's it.
Does anyone have any ideas how I can narrow this down even further? I really feel like it's an ISP issue but my telco says they don't see anything of course as they always do then magically things start working again.
My co-worker says it's probably because of the storm issues on the east coast but I'm in Dallas, and my ISP is Level 3 which has a pop in Dallas as well as my local telco is South Western Bell which is local and the server I'm testing from is local so everything should be staying here.
Thanks...
I have an issue I can't figure out. We have a DS3 connection. Fiber to a MUX in the Data Center then to 3800 series router then to Cisco ASA.
We can get around 41 Mbps download but only 1 to 2 upload at the max and sometimes less than that. This translates to our customers having issues when getting large files from us via FTP or Video.
To narrow it down I've done the following.
Placed a switch between my firewall and my outside internet connection effectively bypassing my firewall and network. I gave myself and outside IP and ran the speed test. The test show the same results of a normal download and very slow upload of not more than 2 Mbps.
I've checked my Internet router for errors and there are virtually no dropped packets, no runts, no overruns, cpu utilization is 10% and memory is 40%. I've also looked at the buffers and the small buffers had maybe 300 misses but that's it.
Does anyone have any ideas how I can narrow this down even further? I really feel like it's an ISP issue but my telco says they don't see anything of course as they always do then magically things start working again.
My co-worker says it's probably because of the storm issues on the east coast but I'm in Dallas, and my ISP is Level 3 which has a pop in Dallas as well as my local telco is South Western Bell which is local and the server I'm testing from is local so everything should be staying here.
Thanks...
Comments
-
Lizano Member Posts: 230 ■■■□□□□□□□With big pipes I have had cpe settings play ugly tricks on me, I've been hooked with a laptop straight to a 50 meg pipe and got poor speeds, change the laptop for a linux box and boom, got 50 megs. That being said:
I would hook up a PC straight to the DS3 router, with nothing else hooked up (I know this is painful cause it basically means you knocked the thing down, but your problem is also painful), try speedtest.net (yes this is not the most reliable but most of the time it's in the right ballpark, I've tested 100 meg metroethernet pipes and gotten 80, 90 megs.), also if you have the ability, use iperf. I have seen where I can't get more than 2-3 megs out of one iperf session, but then I opened several simultaneous iperf sessions and got like 10 sessions each passing 2-3 megs, so the carrier saw me sending 20-30 megs (that was a CPE issue).
I take it your show controllers comes back clean. If your interface stats are clean, and with your network isolated you cant push more than a few megs from a PC after trying web based speed tests, multiple iperf sessions, nor FTP transfers, than I would go back to the carrier and push back, if they still don't see an issue then have them go out and prove through the demarc.