In article <BC8Nd.20417$Vg3.11209@lakeread05>,
Gary <> wrote:

o not mind using 2 PIX's but need to know if it will work before I buy
:another.
:Is your solution a real one that will work or best guess. It looks good to
:me.
I have used a similar solution to deal with overlapping address spaces
for a VPN tunnel:
172.16.N/24 -- remote vpn concentrator <==VPN==> our PIX 172.16/16
On our pix,
name 192.P.Q.R ALocalServer
name 172.16.N.80 RemoteHostIP
name 172.17.N.80 LocalMappedIP
access-list vpn2remote permit ip host ALocalServer host RemoteHostIP
static (outside,inside) RemoteHostIP LocalMappedIP netmak 255.255.255.255 0 0
crypto map vpn-map 1000 match address vpn2remote
When our local server wants to talk to RemoteHostIP (which overlaps one
of our internal networks), we address the packets to LocalMappedIP
instead. The PIX sees the reverse nat [notice the interface order
change], it rewrites the destination IP that we used internally
into the real destination IP and then notices that that the result
matches the vpn tunnel ACL vpn2remote so it pushes the traffic through
the tunnel. It so happens that all of the hosts that RemoteHostIP
needs to talk to on our side are in public IP space, not in the overlapped
IP space [well, it's not really an accident -- we use public IP space
for any host that has to be remotely reachable] so in our situation
there is no issue about having to get the remoted end to do the same
kind of mapping trick. The trick can in theory be done with only
one side doing the mapping, by using both a
static (outside,inside) and a static (inside,outside)
but I haven't tested that.

o not mind using 2 PIX's but need to know if it will work before I buy
:another.
That's what lab tests are for. You do enough PIX work that you should have
an extra PIX 501 or two on hand specifically to be able to test out
scenarios and test out PIX software upgrades and to syntax-check
notable configuration rewrites before putting the revised config
onto a production system.
Keep in mind that entry PIX 501's have that 10 user license, so
you probably would want to go for PIX 506E if you have more than ~20
hosts on each side.
Other thing to keep in mind is that having a PIX or two in the
middle is *not* the same as having a "transparent filter": there are
differences in UDP handling and especially differences in TCP handling
when it comes to long-running connections. And if you have netbios
in there, it really becomes a pain as the PIX doesn't rewrite
addresses that are flowing through TCP 139 [or so the TAC told me
two days ago.]
You also have a problem if the devices on the two "clouds" are on
the same segment without any choke-point inbetween: in such a case,
anyone could get around the filter by simply directly addressing the
remote system instead of addressing the mapped IP corresponding
to the remote system.
If you do have a choke point, then considering that everyone on
one side is going to have to use other IPs to address everyone on
the other side, it probably makes more sense to simply put in one
PIX total and re-number the hosts on one side so they are no longer
in 10.15.0/24.
--
Are we *there* yet??