tech-kern archive

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index][Old Index]

Re: Fast v.s classic forwarding in NetBSD 4.0

Thank you very much for your answer. Explaination is clear, id like to
ask another questions though..

Which of those mechanisms is used by default ? Where to find any
documentation ? Is fast forwarding
fully functional in terms of working with advanced features, like
IPSEC, or are there any known problems
about FF ? Thanks A LOT.


2010/7/12 Joerg Sonnenberger <>:
> On Fri, Jul 09, 2010 at 01:15:58PM +0200, Putrycy wrote:
>> Im having problems understanding relation between "classic"
>> (ip_input.c) and fast (ip_flow.c) forwarding. Could
>> anyone please explain what exactly is the difference, which mechanism
>> is used when, and what is corelation between those two ?
> ip_input.c does a full routing table look up for every packet. It has a
> single entry cache for the most recently used route, but that is
> normally not very efficient.
> ip_flow.c does flow-table based routing. It does a lookup for the packet
> header in an rb-tree and if it finds an entry, it uses the cached route.
> The primary difference between fast forwarding and normal forwarding is
> that the former traverses a bit less of the network stack and is quite a
> bit more efficient if there is a reasonable small to medium set of
> active connections AND the number of routes is high. I doubt it makes a
> different for a home router.
> Joerg

Home | Main Index | Thread Index | Old Index