[#naptabbr]: NAPT
See: network address port translation.
[#nasabbr]: NAS
See: network access storage
[#natabbr]: NAT
See: network address translation
[#naccstor]: Network Access Strorage (“NAT”)

A device on a network which provides files using network protocols such as SMB, NFS, or perhaps HTTP. (See: file transfer, and perhaps especially filesystems provided over networking.) This is distinct from DAS because it involves data over a network, and distinct from a SAN because the data is provided over common networking protocols that are used for transfering files.

[#netprttr]: network address port translation (“NAPT”)

This was a term used by RFC 3022 to refer to a method of translating network traffic by altering the address port numbers. The process seems to be more commonly known by another name: (network) port address translation (“PAT”). (See the section on PAT for more details.)

[#netadrtr]: network address translation (“NAT”)

This is discussed by RFC 3022: Traditional IP Network Address Translator (Traditional NAT), and the section on firewalling: subsection related to NAT.

Some people might consider that the term NAT rightfully applies only to static NAT (“SNAT”) which converts only network addresses, and that any NAT involving ports should be called (network) port address translation (“PAT”). There are a couple of problems with this. First, more authoritively, the “Abstract” section at the top of RFC 3022 discusses this. The practice of translating just the addresses is given the term “Basic Network Address Translation or Basic NAT”. The term “Traditional NAT” refers to being able to use “Basic NAT” and NAPT which is what is also known as PAT. So, NAT does include PAT, officially. Additionally, though, the fact is that PAT often translates the network addresses. One could even argue that a network port is, on a conceptual level, being treated like an address. So, in the end, there seems to be little basis to back up the idea that PAT should not be considered to be NAT.

[#netwrkid]: Network ID

A “network ID” is an identifier that helps to specify a specific network. Each “network” of IP addresses (like, for example, the “192.0.2” class C IPv4 network), including all subnets and supernets, has a “network ID” which looks exactly like the first IP address in the group of addresses.

With VLSM and CIDR permitting subnets and supernets, a “network ID” is no longer sufficient to identify a specific network. Some sort of indication of a subnet size, such as a CIDR-style prefix length, is also needed.

Since the “network ID” of a network looks just like the first address of that network, people also commonly call the first address a “network ID”.

With IPv4, the first address of each subnet is generally considered to be “unusable” for unicast traffic. (Also, the broadcast address is generally considered to be “unusable” for unicast traffic.) Using this sort of number as an identifier of a network is shown in the paragraph above section 7.1 of RFC 919: Broadcasting Internet Datagrams: section 7: Proposed Standards for Broadcast IP Addressing.

While it may be clear why an IPv4 broadcast address is generally unusable for unicast traffic (because the address is used for broadcasting), the logic that prohibits the ability to use a network ID is not as widely discussed. The basic reason why these addresses are unusable is traced back to some hosts that used the first address (or specifically an address ending with a certain number of bits that are all cleared to zero) as being a broadcast address. The “Trailer Formats” section of RFC 894 (“A Standard for the Transmission of IP Datagrams over Ethernet Networks”) page 2 discusses this a bit. Section 3.3.6 of RFC 1112 (page 66) (and the next page) does identify “4.2BSD Unix and its derivatives, but not 4.3BSD” as being an example of “a class of hosts” “that use non-standard broadcast address forms, substituting 0 for -1.” (The references to “-1” in that RFC refer to a certain amount of bits which are all set to a value of one. This may be noted by RFC 1122 page 30.) The RFC goes on, with RFC 1122 page 67, to say “A host MAY optionally have a configuration option to choose the 0 or the -1 form of the broadcast address” ... but “SHOULD default to the standard (-1) form.” That is really just referring to outgoing communications. The sentence split between pages 66 and 67 states, “All hosts SHOULD recognize and accept any of these non-standard broadcast addresses as the destination address of an incoming datagram.” RFC 3021 section 2.2 says “The {0, 0} form of a limited broadcast is obsolete, but may still be present in a network.”

Therefore, if the goal in mind is successful network communications, or if trying to sound compliant with common standards, then the generally recommended approach is to avoid the active use of the first address per subnet. Designing the network in a way that conflicts with a design that is allowed by an RFC is a generally bad practice, and the design of using the network ID as a broadcast address is specifically permitted by RFC 1122 page 67.

In reality, the commonly supported broadcast address is now the final address of a subnet, but a lot of training material will instruct students to avoid trying to use the “Network ID” as if it was unusable. This limiting rule, of not using the “Network ID” address (however obsolete/unnecessary the rule may currently be), may be enforced well enough to warrant the non-usage of these sort of addresses. Attempting to use the addresses may cause some problems, by at least some devices and/or software. Even if problems are not visible during initial testing, changes in a network could lead to some software that assumes these addresses are not used by a single device. (Such a device might be treating the first address as generally unusable, or might be trying to use old behavior of using those addresses for another purpose, such as using the address as a broadcast address.) Such a device might be able to communicate with some other devices, so this sort of issue could be fairly challenging to determine the actual problem.

Although the Network ID address may commonly be unused with IPv4 due to what may be a fairly obscure reference in an old RFC, such support is quite clear and specific with IPv6's specifications where this type of address is given a different name: the “Subnet-Router anycast address”. (A description of this IPv6 “Subnet-Router anycast address” is described by RFC 4291 section 2.6.1).

[#nos]: Network OS (“NOS”)

The term “Network OS” may vary a bit. It could simply refer to an operating system that contains code to be able to participate on a network (communicating with other devices). Or, the term may refer to an operating system that contains server software that provides services that are likely to be utilized by other devices on a network.

[#nic]: NIC

Originally, and still officially, this has been given the meaning of “network interface card”. However, in practice, networking communication is often implemented now without requiring a separate add-on card to be added to a computer. The term NIC still applies, even if all it is really referring to is embedded “network-capable integrated circuitry” components. Any Ethernet port may properly be called a NIC, so the phrase “network interface card”, which is what the term NIC officially stands for, is now often a misnomer.

See also: communications hardware: NIC.

[#node]: node

There are a few definitions of the word “node”, as shown by Wikipedia's (disambiguation) page for the word “node”. Perhaps a rather common trait of a “node” is that a “node” is capable of acting like an end-point (rather than being a piece of centralized infrastructure).

[#nodenet]: node (of a network)

There are some different definitions.

IP nodes

To keep things simple: a device that has an IP address can be considered to be a node on the IP network. If the node can forward packets, it is a router; otherwise the node is called a host.

  • An official source that provides a definition is RFC 1883 (IPv6) section 2: “Terminology”, as well as updates to that document (like RFC 2460 section 2). The definition given there is “a device that implements IPv6”. However, the document giving that definition is a document that is all about the specific topic of IPv6.
  • The term “node” is certainly not IPv6 specific. That's clearly true because, as a generalization, it is entirely appropriate to also use the term to refer to any device that communicates using IPv4.
  • When discussing networks that implement the Internet Protocol, RFC 4861 (NDP for IPv6), section 2.1 (General Terminology) defines a “node” as “a device that implements IP.” (Unusally, that document generally uses the term “IP” to refer specifically to “IPv6”, as discussed earlier in that section 2.1.)
  • RFC 588: “London node is now up” describes a news event from October 1973, and describes a node of the ARPANET.
  • RFC 1933 section 1.2 describes different “Types of Nodes”. Every node type starts with the phrase, “A host or router that implements IP”.
  • RFC 1034 section 2.4: Elements of the DNS refers to a “node and leaf of the domain name space tree”.

IP nodes are also hosts. Glossary entry for “host” provides some additional references for that similar term.

A “dumb switch” may be able to forward traffic to other devices within a (sub)network, but does not have a network address to be able to receive network traffic intended for itself. Since the switch may be operating at only OSI Model Layer 2, and relying just on MAC addresses, the device may not be using IP at all. So, does that mean that the switch is not a node?

Well, at this point, things can get a bit trickier. Wikipedia's article on “Node (networking)”: section on “Computer network nodes” says, “a physical network node may... be a... switch” Yet the same section says, just briefly later, “some datalink layer devices such as switches... do not have an IP host address... and are not considered to be Internet nodes or hosts, but as physical network nodes and LAN nodes.” So, in this case, the switch is not an “Internet node” but it is a “physical network node”. What this clearly means is that the term “node” can be getting used in different ways, at the same time. A diagram of all “IP nodes” would not include the switch, but a diagram of all “physical nodes” would. The switch would fit the description of one type of node, but not the other type of node.

In computer networking, one definition of the word “node” is any computer, or other device, which is connected to the network.