Steve said:
We develop application that communicates to certain devices over TCP\IP
network.
We want to test how our app communicates with multiple 40-50 devices
simultaneously. It is impossible to set up so many real devices in our ofiice.
I wonder is there any way to simulate connections to 50 fake IP addresses
while actually communicating to one real IP address. In ideal case we would
like setup ip forwarding table which for example forward all IP packets
destined to fake IP addresses 172.18.7.1 - 50 to real IP address 172.16.7.99.
Is it possible?
Maybe,..but I hope not.
If yes, how can configure this IP forwarding?
That is not what IP Forwarding is for. Technically "IP Forwarding" is just
normal Layer3 routing. Look in the TCP/IP properties of any old NT4.0 box
(even NT Workstation) and you will see it listed there as an option to turn
on or off. SOHO Devices are famous for abusing and misusing terminology in
their documentation if that is where you got the term from. What you
probably really have in mind is called Static NAT or One-to-One NAT
depending on exactly what is done. In any case,..that still is not what you
want to do.
The first question is what does the "App" connect "to". It cannot connect to
a machine simply because it exists,..there actually has to be something
running on the machine that interfaces with the "App".
You could put multiple IP#s on one Nic of a Client, but that is not an
acuarte test because there is still only one "instance" of the Client
Application that it interfaces with the Server "App". It will also always
reply back to the Server App on the Default IP# and not the IP# it was
received on. The communication also has to share the same wire between the
Client and the first Switch/Hub. So this method fails in three ways.
You could run multiple OS's using MS Virtual PC or MS Virtual Server but
that is not an acuarte test because the hardware is "shared" and each
instance of a Virtual PC slows the hardware down and the slow response from
the client caused by that invalidates the test.
The right way, and accuarte way to test this is to actually have 50 clients
using the proper level of hardware you expect to be used in production for
it to connect to.
If you don't have the environment to do that, then maybe you can "beta" the
App out to someone who can test it for you.
