PAT vs NAT

So I seem to be having some confusion here...

I get the difference between PAT and NAT, but Juniper seems to over complicate where/when you would use each. Obviously NAT is for one to one, one to many, etc.....but in what instance would you use PAT?

Find more posts tagged with

Comments

networker050184

NAT is one to one. PAT is many to one.

NightShade03

networker050184 wrote: »

NAT is one to one. PAT is many to one.

Hmmm maybe my Nat/pat understanding is off...thanks for the reply though

Zartanasaurus

NightShade03 wrote: »

Hmmm maybe my Nat/pat understanding is off...thanks for the reply though

It's confusing b/c everyone calls it NAT when they usually mean PAT.

There's Static NAT, Dynamic NAT and PAT (or NAT Overloading).

Static NAT is a one to one mapping. Each private host consumes one external IP.

Dynamic NAT is kind of like DHCP. You configure a pool of public addresses. Private hosts make a connection and the router hands out the first available public IP. Once the pool is filled, no more private hosts can be translated. When a connection goes stale, an IP is returned to the pool.

PAT is what most people mean when they say NAT. Each private connection to the outside is mapped to a public IP + port combination. So even with one public IP, you could have several private hosts using the same public IP.

Static = one to one
Dynamic = many to many on a one to one basis
PAT = one to many

Most likely, you'd end up doing static (an internet facing server) or PAT (a group of client machines on a subnet).

NightShade03

Excellent clarification!! I noticed that Cisco refers to PAT as "overloading" which is probably why Juniper calling it PAT is confusing me. Makes a lot more sense now, just have figure out the Juniper specifics for configuration. Thanks.