>[оверквотинг удален]
>> железо  должно и большую нагрузку ???
>> Подскажите?? ПЖ
> Вот для общего развития
> http://www.cisco.com/en/US/products/hw/routers/ps133/product...
> А вот вырезка по поводу фрагментации
> Reassemblies can drive up the CPU very high if the CPU has
> to reassemble a large number of packets.
> For troubleshooting high CPU utilization due to fragmentation, issue the tcp mss-adjust
> 1400 command on the interface which sets the maximum segment size
> (MSS) value of TCP synchronize/start (SYN) packets going through a router.

вот что есть на show logging
4777: Apr 22 02:18:06.338: %IP_VFR-4-FRAG_TABLE_OVERFLOW: Virtual-Access82: the fragment table has reached its maximum threshold 16
2544778: Apr 22 02:21:00.444: %IP_VFR-4-FRAG_TABLE_OVERFLOW: Virtual-Access82: the fragment table has reached its maximum threshold 16
2544779: Apr 22 02:28:35.186: %IP_VFR-4-FRAG_TABLE_OVERFLOW: Virtual-Access82: the fragment table has reached its maximum threshold 16
2544780: Apr 22 02:29:05.716: %IP_VFR-4-FRAG_TABLE_OVERFLOW: Virtual-Access82: the fragment table has reached its maximum threshold
и так для ещё десяток сессий

поднимаються они по шаблону
nterface Virtual-Template1
ip unnumbered GigabitEthernet0/2.801
ip access-group acl-user-in in
ip access-group acl-user-out out
no ip redirects
no ip unreachables
no ip proxy-arp
ip mtu 1492
ip flow ingress
ip flow egress
ip nat inside
ip virtual-reassembly
no logging event link-status
autodetect encapsulation ppp
ppp authentication chap callin
я так понимаю что ip tcp adjust-mss <500-1460> именно сюда и надо ?!