|
Hello,
On Wed, 16 Jan 2002, Roberto Nibali wrote:
> personally work on reducing network latency just for you :). I think
> your tomcat processes will die before that. And I also vote for the
> third solution, although I would modify it to use LVS_DR. In your
> depicted solution you favour LVS_NAT. OTHO several people on the list
> have stated that NAT seems to be almost as fast as DR. YMMV, I just gave
Yes, many people think that the actual packet mangling is an evil
in the NAT processing. The picture is different: the NAT processing in
2.2 uses 2 input routing calls instead of one and this totally kills
the forwarding of packets from/to many destinations. Such problems
are mostly caused from the bad hash function used from the routing code
and from the fact that the routing cache has hard limit for entries.
Of course, the NAT setups handle more traffic than the other ones,
so this can be a good reason to avoid them when the director's power
is not enough to handle so many packets. So, in 2.4 the difference
between the DR and NAT processing in out->in direction can not be
noticed (at least in my tests) because only one route call is used,
for all methods.
> Best regards,
> Roberto Nibali, ratz
Regards
--
Julian Anastasov <ja@xxxxxx>
|
