How to fix loss on Cisco interface?

Question

Computer networks

A few days after connecting an additional provider to the Cisco 2821, terrible losses began on all external IPs on this interface. I'm not sure if it's because of the provider, but still.
Wiring diagram
ISP1 -> Switch Untag Vlan 668 (11 port) -> Tag port 48 -> GE0/0.668
ISP2 -> Switch Untag Vlan 700 (12 port) -> Tag port 48 -> GE0/0.700
...
Problem with queues on the interface (IHQ/IQD).

vpn#sh interfaces summary 

 *: interface is up
 IHQ: pkts in input hold queue     IQD: pkts dropped from input queue
 OHQ: pkts in output hold queue    OQD: pkts dropped from output queue
 RXBS: rx rate (bits/sec)          RXPS: rx rate (pkts/sec)
 TXBS: tx rate (bits/sec)          TXPS: tx rate (pkts/sec)
 TRTL: throttle count

  Interface                   IHQ       IQD       OHQ       OQD      RXBS      RXPS      TXBS      TXPS      TRTL
-----------------------------------------------------------------------------------------------------------------
  Em0/0                         0         0         0         0         0         0         0         0         0
* GigabitEthernet0/0           81   1345261         0         0   3012000       349    350000       329   1037935
  Gi0/0.667                     -         -         -         -         -         -         -         -         -
* Gi0/0.668                     -         -         -         -         -         -         -         -         -
* Gi0/0.700                     -         -         -         -         -         -         -         -         -
* GigabitEthernet0/1            0      1343         0         0   4381000      1775   7015000      1768       365
* Gi0/1.100                     -         -         -         -         -         -         -         -         -

After applying 'hold-queue 4096 in' on the interface, the pings stabilized, but after half an hour it was no longer possible to connect via ssh (as I understand it, the problem of outgoing and incoming queue mismatch) but this is the tenth thing, I would like to understand the root cause.

Answer 1

1) While observing the problem, reset the counters on the interfaces (offhand clear counters), after ten minutes, request in the console show interface gi0/0and provide the output.
2) While observing the problem, query the console show processes cpu sorted 1minand provide the output.
3) On the switch, configure SPAN/mirror traffic from both VLANs, compare the traffic in production mode and when observing the problem. Pay attention to the traffic level (bps, pps), broadcast, multicast traffic; traffic addressed directly to the router.

Answer 2

Are you ok with physics? What's there? Optics, copper, pigeons carry packages?