Speedfusion, MTU, packet fragments, cisco ios virtual-reassembly


#1

Hi all –

I’m wondering about speedfusion behavior with regard to fragmenting packets. Does the peplink send packets larger than the WAN interface’s MTU which it does layer-3 packet fragmenting on?

Does the peplink expect the ability to send fragmented ip/udp/tcp packets past to its next-hop interface? The next-hop after the speedfusion on one of my wan’s is a cisco router which has a serial interface connected to the layer-2 WAN device. This cisco router is currently configured to drop fragmented packets – in ios-land this is done by applying the ip virtual-reassembly drop-fragments command to the relevant cisco interfaces. With this configuration, the cisco virtual-reassembly feature will simply drop any fragmented packets coming into it.

Should I disable this fragment filtering?

More generally, are there recommendations regarding the use or non-use of the cisco ‘virtual fragment reassembly’ feature? Prior to migrating to the peplink, we had been using the virtual fragment reassembly feature on the cisco – not as a fragment filter but doing actual fragment reassembly with ‘ip virtual-reassembly.’ This ios feature reassembles fragmented packets before forwarding them along, it was needed because we had a fine-grained firewall defined in our cisco router – and this was the best/easiest way to get correct packet filtering behavior of fragmented packets.

Many thanks,


#2

I should comment that the problem was identified (ISP issue upstream) – and not related to any of the above mentioned questions. For now I’m leaving the fragment filtering in place. No urgency, but my curiosity wouldn’t mind any recommendations if someone has any.

Thanks,