WanDSL Saturation

[Fotiosmark]

Hello,

I have a very weird problem.

There are 10 VPN site to site which are working fine. The issue I am facing, and I really looked everywhere, is that 3 of those sites are having issues with the Wan line dsl.

The thing is when the high latency begins and disconnects start to happen from the WAN, it starts at all the 3 sites at the same time, which are Geographicaly away from each other!!!

I don't know if that has something to do with the VPN or the Provider. Would anyone have any ideas?

There nothing really on the logs.

The show dsl interf atm 0 give me the below

It happens random within the day, sometimes even before the offices open!!!

----#sh dsl interface atm0
ATM0
Alcatel 20190 chipset information
                ATU-R (DS)                      ATU-C (US)
Modem Status:    Showtime (DMTDSL_SHOWTIME)
DSL Mode:        ITU G.992.5 (ADSL2+) Annex B
ITU STD NUM:     0x03                            0x2
Chip Vendor ID:  'STMI'                          'BDCM'
Chip Vendor Specific:  0x0000                    0x939E
Chip Vendor Country:   0x0F                      0xB5
Modem Vendor ID: 'CSCO'                          'BDCM'
Modem Vendor Specific: 0x0000                    0x0000
Modem Vendor Country:  0xB5                      0xB5
Serial Number Near:    FHK1246248P
Serial Number Far:     
Modem VerChip ID:        C196 (0)
DFE BOM:         DFE3.0 Annex B (2)
Capacity Used:   98%                             100%
Noise Margin:    10.5 dB                         13.0 dB
Output Power:    18.5 dBm                        12.5 dBm
Attenuation:     13.0 dB                          8.0 dB
FEC ES Errors:    0                              1103
ES Errors:        1                              265
SES Errors:       1                              90
LOSES Errors:     1                               0
UES Errors:       0                              182405
Defect Status:   None                            None                        
Last Fail Code:  None
Watchdog Counter: 0x65
Watchdog Resets: 0
Selftest Result: 0x00
Subfunction:     0x00 (UR2)
Interrupts:      5781 (0 spurious)
PHY Access Err:  0
Activations:     1
LED Status:      ON
LED On Time:     100
LED Off Time:    100
Init FW:         init_AMR-3.0.014_no_bist.bin
Operation FW:    AMR-3.0.014.bin
FW Source:       embedded
FW Version:      3.0.14

                 DS Channel1      DS Channel0   US Channel1       US Channel0
Speed (kbps):             0            11264             0               861
Cells:                    0         21579116             0         166737639
Reed-Solomon EC:          0              340             0              8303
CRC Errors:               0               34             0               844
Header Errors:            0               35             0             17938
Total BER:                0E-0           3994E-11
Leakage Average BER:      0E-0           6359E-11
Interleave Delay:         0               29             0                58
                        ATU-R (DS)      ATU-C (US)
Bitswap:               enabled            enabled
Bitswap success:          0                   0
Bitswap failure:          0                   0

[Georg Pauwen]

Hello,

what do the router logs at the affected sites show ? Are the configs and the hardware the same at the sites that work and that do not work ?

[Fotiosmark]

Hi Georg,
Thanks for the reply.
Those sites have cisco876 adsl with PPPoe configured. What I did, thinking it might be a routing issue or something from the Provider, (all sites have static IP obviously) I configured a cisco with the same username/password in the dialer, told one site to shut it down which had issues, and plug it to another adsl at my central Site
It seems that again, all sites went through the roof with high latency Saturday (offices were closed) but the one at my Central it was fine.
I am thinking it must be Noise or Attenuation since those problematic remote sites are in very remote areas.
One output is this
Noise Margin: 10.0 dB 13.0 dB
Output Power: 18.5 dBm 12.5 dBm
Attenuation: 13.0 dB 8.0 dB
In my area, Noise Margin below 12dB are problematic and it has to do with ISP.
Would you think that is the problem?

[Fotiosmark]

as for the logs, the only thing that it shows sometimes are invalid SPIs within IPsec and I think that occur due to latency

[Joseph W. Doherty · Accepted solution]

Latency often denotes congestion somewhere also the path. If it's happening at the same time to sites supported by the same provider, it could be there's congestion somewhere in the provider's network.

If your hub supports SLA, and your spokes support a SLA responder, an on-going traceroute SLA test might catch a provider congestion point.

[Fotiosmark]

well,
They started chaning their connections from adsl to vdsl... :) Once thats dones, its fixed lol