NAT Concepts
Millions of private addresses live behind a handful of public ones. The mechanism that makes that possible is Network Address Translation (NAT) — a router rewriting IP addresses as packets cross its boundaries. This lesson builds the vocabulary you need before touching a single command: public versus private addressing, why NAT exists, and the four-way inside/outside, local/global terminology that the exam loves.
Public vs private addressing
Every interface on the Internet needs an address that is globally unique so packets can be routed to exactly one destination. Such an address — assigned by an ISP or an address registry, and advertised in the public routing tables — is a public (or global) address. Public addresses are a scarce resource: IPv4 address space is only 32 bits, and the last blocks were allocated from the registries years ago.
The response to scarcity was private addressing. RFC 1918 set aside three ranges that are not routed on the public Internet and may be reused freely inside any organisation:
| Range | Prefix | Hosts per network (max) | Typical use |
|---|---|---|---|
| 10.0.0.0 – 10.255.255.255 | /8 | 16,777,214 | Large enterprise networks |
| 172.16.0.0 – 172.31.255.255 | /12 | 1,048,574 | Mid-size networks |
| 192.168.0.0 – 192.168.255.255 | /16 | 65,534 | Small office / home |
Remember the other reserved blocks that are often named in the same breath: APIPA / link-local 169.254.0.0/16 (self-assigned when no DHCP lease can be reached), CGN (carrier-grade NAT) 100.64.0.0/10 used by ISPs for their own subscribers, and the documentation addresses 192.0.2.0/24, 198.51.100.0/24, and 203.0.113.0/24 (RFC 5737) — used in textbooks and examples precisely so they are never confused with real public addresses.
Why NAT exists
NAT (Network Address Translation) is the process of rewriting the IP address (and often the TCP/UDP port) in a packet header as it passes through a router. The classic driver is simple: an organisation may hold thousands of hosts addressed from RFC 1918 space, but only one or a few public addresses. NAT maps the private addresses to the public ones at the edge.
Two forces made NAT essential. First, address conservation: the number of connected devices far exceeds the IPv4 space (the original design assumption of a unique address per device is long gone). Second, renumbering avoidance: an organisation can keep its internal numbering stable (even overlapping another company’s, such as two businesses that merge) and let the edge router bridge to the public world.
NAT also delivers a side benefit that is really a security property: internal addresses are not visible from outside, so external hosts cannot directly initiate connections into the internal network unless an explicit translation exists. Do not call NAT a firewall — it is not a security device — but the exam expects you to know the masking effect.
The four addresses: inside/outside, local/global
This is the single most misunderstood part of NAT and a favourite theme of the CCNA exam. NAT uses two physical views (inside versus outside) and two visibility views (local versus global) — giving exactly four terms. Keep the definitions attached to the router’s point of view:
- Inside local — the address of an inside host as seen from the inside network. This is the real, private address the host was configured with (for example 10.0.0.5).
- Inside global — the address of that same inside host as seen from the outside network. This is the public address the translator maps it to (for example 200.100.10.5).
- Outside local — the address of an outside host as seen from the inside network. Normally this equals the real public address of the remote server (for example 8.8.8.8).
- Outside global — the address of that same outside host as seen from the outside network. For ordinary outbound NAT, outside local and outside global are the same because you are not translating the far end.
In the lab and in show ip nat translations, the table always lists four columns in the order Inside global, Inside local, Outside local, Outside global. Practise reading one row now: --- 200.100.10.5 10.0.0.5 --- --- means inside host 10.0.0.5 is presented to the outside as 200.100.10.5. The far end was not translated, so the outside columns are empty.
Inside and outside interfaces
Before NAT can decide anything, it must know which interfaces are inside and which are outside. Every packet is then examined for translations in both directions. In IOS you declare this per interface:
R1(config)# interface gigabitEthernet 0/0 R1(config-if)# ip nat inside R1(config-if)# exit R1(config)# interface gigabitEthernet 0/1 R1(config-if)# ip nat outside
Inside hosts receive private addresses via DHCP or static configuration, and outside hosts are anything reachable beyond the edge interface — typically the ISP link. A common exam scenario gives you exactly this topology: LAN (inside) → NAT router → public Internet/SERVER (outside). Notice that ip nat inside and ip nat outside do not create any translation by themselves; they only mark which side is which so that the actual ip nat inside source ... statements (Lessons 8 and 9) know what to translate.
Static vs dynamic vs PAT
Three ways to run NAT answer three different questions. The exam expects you to distinguish them on two axes: is the mapping fixed or temporary? and how many public addresses do you need?
| Type | Mapping | 1:1 or many:1? | Public addresses needed | Typical use |
|---|---|---|---|---|
| Static NAT | Manually configured, never expires; the same inside local always maps to the same inside global. | 1:1 | One public address per host | A public web/email server that must be reachable from the Internet. |
| Dynamic NAT | Automatic, temporary; an inside host grabs any free address from a pool of public addresses. | 1:1 while it lives | One public address per concurrent host | A group of hosts needing outbound access; the pool must be big enough. |
| PAT (overload) | Automatic, temporary; many inside hosts share one public address, distinguished by unique source port numbers. | many:1 | One (or a few) public address(es) | Almost every home and small-office router on the planet. |
overload keyword — is NAT with multiplexing. Because TCP and UDP sessions can be told apart by their source port number, thousands of private hosts can share a single public address at once. The router tracks which private host owns which (public IP + port) combination.The key mental model: static and dynamic NAT are both 1:1 (they change only the IP address), whereas PAT is many:1 (it changes both address and port). A pool that is too small, or a static setup with too few addresses, means hosts simply cannot get a translation when the pool is exhausted — a classic dynamic-NAT failure mode that PAT avoids.
Benefits and limitations of NAT
NAT conserves public IPv4 address space and hides the internal numbering — but it is not free. Pick a side to see the arguments.
NAT also interacts directly with other IP Services skills from this week: NAT traversal matters for FTP (recall passive mode from Lesson 6, which exists precisely because active mode dies behind NAT), and the time synchronisation you configure in Lesson 10 must reference reachable addresses — which, inside a NATed network, means addressing the translation correctly.
Review
Check what you have learned with these review questions.
Sources & references
- Rekhter, Y. et al. RFC 1918: Address Allocation for Private Internets (1996). rfc-editor.org/rfc/rfc1918.
- Arkko, J. & Cotton, M. RFC 5737: IPv4 Address Blocks Reserved for Documentation (2010). rfc-editor.org/rfc/rfc5737.
- Srisuresh, P. & Egevang, K. RFC 3022: Traditional IP Network Address Translator (Traditional NAT) (2001). rfc-editor.org/rfc/rfc3022.
- IANA. Special-Purpose IP Address Registries. iana.org.
- Cisco. CCNA (200-301) Exam Topics, Domain 4.0 IP Services (NAT). cisco.com.
Target exams: CCNA 200-301 v1.1 · CyberOps Associate 200-201. Last content verification: 2026-10-11.