Communications Hardware

Endpoint connectivity
Connectivity for computers/workstations
[#nic]: Network interfaces

Network interfaces are often referred to as a “NIC” (or even a non-capitalized abbreviation, “nic”), which stands for “network interface card”. This is true even for network ports of circuitry built onto a motherboard (which doesn't use a card) and for wireless solutions (which may not have a port for a cable to be plugged into).

See also: Glossary: NIC.

link light

See: link light, a visual representation of whether there is a connection, similar to what gets reported by media sense.

[#potsconn]: POTS

Way back in the day, computers could use a DE-9 serial cable to connect to an external modem. Later, modems were sold in the form of expansion cards. Some laptops have supported an internal modem. However, internal support for such dial-up service has waned because of the spread of Internet access that is substantially faster to what is provided by POTS. See: dial-up modems for further discussion about using POTS.

Network icons

Cisco's network infrastructure icons package has an icon named “modem.jpg” (which is also mentioned, and is shown, the section about broadband connectivity).

There are more icons related to telephones and/or voice communication. For instance, there's also a some “pbx*.jpg” files. There's also “fax.jpg” which basically looks rather like a printer (but different than the “printer.jpg” because “fax.jpg” has a place for paper to be placed). Other icons may represent telephones or products that are identified as supporting voice communication.

[#comhwprt]: Communications ports

Using a “null modem” cable was an option involving using communications ports. (The word “was” is used because modern computers do not typically come with such ports, although USB adapters might be available.) So, this is not something typically done on modern networks; this is mainly being mentioned as a historical footnote. Using such ports often involved software designed to communicate over such ports, and could allow for transferring files or transmitting raw text. Details about that are in the section on being a “dumb terminal”. Additionally, information has been mentioned in other sections of the website such as the section about COM ports (used by hardware “serial communication ports”) and LPT ports (used by hardware “parallel ports”). Users of DOS and compatible platforms (like Microsoft Windows) may find some software at TOOGAM's Terminal Software. Users of Unix could use minicom or picocom or tip. Some additional specialized software has been available, such as Interlnk (and its companion program, Intersvr) which came bundled with MS-DOS 6.

Eventually, TCP/IPv4 started to become important, and these methods of simpler raw communication became less popular. In the mid 1990's, some people would log into a provider and use PPP or SLIP to enable some TCP/IPv4 connectivity. Related/similar terms could include CSLIP or SLiRP for Unix.

OpenBSD FAQ 7: section on serial cables has this to say:

Note that serial interfacing is NOT a trivial task -- you will often need unusual cables, and ports are not standardized between machines, in some cases, not even consistent on one machine. It is assumed you know how to select the appropriate cable to go between your computer and the device acting as your serial terminal. A full tutorial on serial interfacing is beyond the scope of this article, however, we offer one hint: just because the ends plug in doesn't mean it will work.

Some related information may actually be found in the glossary: see:

Often, the electrical signaling used for a serial port was the RS-232 standard. References to an RS232 port may frequently be referring to a serial port, and perhaps most commonly a DE-9 port.

Parallel ports became less popular when USB ports offered superior speed. Before that happened, an “Enhanced Parallel Port” (“EPP”) functionality offered increased speeds, up to 2MB/sec, and less CPU overhead. “Extended Capability Port” (“ECP”) offered 2.5MB/sec and even less overhead, although it did require using up a DMA channel. (IEEE 1284 modes) Wikipedia's article for “PC System Design Guide”: “Color-coding scheme for connectors and ports” section identifies PC99 as the source for the color standardization that started to make ports be “Burgundy” (“purple”/“maroon”). These purple ports are usually new enough to support the IEEE 1284 modes (EPP or the newer ECP).

A paper book by Sybex (about the LPIC-1 certification for Linux), on page 118, stated:

“Older external devices, such as parallel and RS-232 ports, are officially coldplug in nature. In practice, many people treat these devices as they were hotplug, and they can usually get away with it; but there is a risk of damage, so it's safest to power down a com-
puter before connecting or disconnecting such a device.”
Site connectivity
Wired communications methods
Broadband connections
Cable modem, DSL modem
Files from Cisco's
network icons package:

broadband router.jpg

cable modem.jpg

(also used for POTS)

These wires (telephone for DSL, and wiring for cable television programming) were designed for the purpose of delivering information.

When broadband was first coming out, there was a such thing as an add-on card that would get plugged straight into a computer's motherboard, and which had a connector for the type of broadband (such as cable). Fortunately, the standard has now been to have this connectivity come in the form of an external box. This is fortunate because now the connectivity for a computer is less dedicated, so a computer can plug into a standardized switch and use that same connection to communicate to other devices on the local network.

Some broadband modems come with some capabilities such as providing some settings for firewalling unwanted traffic. Some may come with an antenna, to be able to provide wireless access. In theory there is no big problem with this approach, although the capabilities of such devices may inferior to more specialized dedicated devices. Also, if the broadband modem is damaged due to an unprotected connection to wires that experience an electrical surge, the fried device may break. In that case, it is nicer if an internal network's connectivity is not broken until the Internet access gets fixed. So, combining capabilities into one device does form a “single point of failure” for multiple pieces of functionality.

[#bitonpwr]: Data sent over electrical power wiring

Some research has gone into providing networking over electrical power. This has not been deployed very widely in the United States of America. However, the option may seem more attractive in some other nations which have built up less infrastructure for telephone service and cable television content delivery.

The potential for causing electrical interference, which may affect other technologies, may be noteworthy. This issue alone may be a cause for severely stalling deployment of the technology. Wikipedia's article for “Broadband over power lines”, section called “Smart grids and use of BPL by power companies” notes, “there is no one single compelling reason to carry data on the existing power lines” ... “except in remote regions where fibre optic networks would not normally be deployed at all. Power network architectures with many transformers are more likely to be served using fibre.”

American Radio Relay League page about Broadband over Power Lines (“BPL”) notes, “Because power lines are not designed to prevent radiation of RF energy, BPL represents a significant potential interference source for all radio services using this frequency range” ... “Overhead electrical power lines and residential wiring act as antennas and overhead power lines radiate the broadband signals as radio signals throughout entire neighborhoods and along roadsides. Interference has been observed nearly one mile from the nearest access BPL source.” “the power levels typically used by BPL systems are approximately 30 to 40 dB greater than the FCC limits for other unintentional emitters. To put this into layman's terms, one BPL device generates as much noise as 1000 to 10,000 other devices”

The article goes on to discuss using spectral masks, a process that the article calls “notching”, as being able to resolve the problem. However, it does not appear that notching has been widespread. So, although a viable technology may exist, the technology has not been deployed.

American Radio Relay League page about Broadband over Power Lines (“BPL”) goes on to state, “So far, access BPL has been deployed in numerous temporary test sites but in few commercial installations. The vast majority of these trials were abandoned by the electric utilities or BPL manufacturers involved, for various reasons, and most commercial installations have been shut down because they proved to be not economically viable. It must be noted that a number of these installations were documented as causing harmful interference” ... “that were not resolved prior to their being shut down.” Wikipedia's article on “Broadband over power lines” notes some deployments that have been shut down.

See also: Wikipedia's article for “Power line communication”: section called “Broadband over power line (BPL)”. Other sections of that Wikipedia page, such as Wikipedia's page for “Power line communication”, section called “Home networking (LAN)”, may discuss similar topics.

Sending bits over electrical wiring within a building

Creating LANs by using power lines is something that can happen, although there have been reported difficulties (such as half of the outlets (generally, “every other” outlet) being unable to communicate with the other half?) Data speeds have often not been substantially better than what is achievable with Wi-Fi.

[#potsmodm]: Dial-up modems for a “public switched telephone network” (“PSTN”) service

from Cisco's
network icons

PSTN has also become known as “Plain ol' telephone service” (“POTS”).

Using dial-up modems is generally not recommended due to their slow speed. Even basic protocols such as DNS and the creation of TCP connections can seem pretty slow, and web-surfing may be fairly slow even when visiting sites with a few basic graphics and quite a bit of text. Media-rich sites are probably not going to be a pleasant experience. However, there may be some scenarios where such equipment could potentially be somehow useful.

There may be some ways to provide dial-up modem speed increases.

(Further explanation should be provided, about how to use the commands. These details may eventually be found in a referenced section that describes the communications software that gets used. Currently, there is no such text being provided.) (being a dumb terminal may provide names of some such software.)

To dial faster, use:


...or play with that value a bit (setting the S11 register to a value lower than even 40). The value might be decreasable without incident (resulting in a smaller speed improvement), although values that get too low (e.g. 25) might not work everytime (or even anytime) when using some equipment, including phone company infrastructure.

Perhaps such modems are most commonly used for three things: connecting to the Internet, using terminal software as users interact with a remote system, and transfering files. For the current time, details about TOOGAM's software archive: Terminal software (section about “Serial” communications software) and TOOGAM's Software Archive: File transfering software may be more detailed resources.

Here is some more information about using dial-up modems:

Consider using ATM1L3D for the loudest possible sound, or ATM1L0 for a quieter sound. (The “D at the end of the first example, which was just mentioned, specifies to start dialisting after setting the audio to on (M1) and setting the loudness level to 3 out of 3 (L3. Using ATM0 could result in turning off the sound. That might be nice for those who don't want to hear the noise of modems communicating, and may be ideal for automated systems where people are not closely monitoring the connection. However, since it also disables peoples' ability to hear a “busy signal”, using ATM1L0 or ATM1L1 may be better.
Other commands

Enabling features such as Caller ID, “Modem on Hold” (“MOH”) which can allow a signal for “call waiting” to occur on a phone line without the phone line being very likely to drop the connection, enabling faxes, and possibly supporting voice calls (e.g. voicemail).

The Official DOOM FAQ: Section 4 (subsection “[19-3]”) has some initialization strings. (Since that game only supported communications that were 9600bps and faster, the section tended to document a fair number of 14.4kbps modems, while not having details about 2400bps or slower modems). In general, ATZ was the Hayes-compatible standard method of resetting the modem to a particular set of default settings, which might be customizable (depending on the modem being used). Using the Z might disable all later options, or even all options, and this might cause the modem to initialize, including hanging up on any active connection.

Using X3 in a dial string could cause the modem to initiate “blind dialing”, which means it would attempt to dial even before it recognizes that there is a functioning dial tone. Many modems could support a format of ATDT765-4321R to try to dial and then, upon noticing a connection on the phone line, initiating sound similar to what is typically done when the modem tries to answer a phone line. Combining that with a modem that blind dials after a certain number of rings might allow the modems to connect after another device, such as an answering machine, picked up the line. Though rarely implemented, this could potentially allow a phone line to be shared between an answering machine and a modem: Both devices would attempt to respond to an incoming call, and the modem would just have no effect unless the incoming call was another modem dialing with an R after the phone number.

ATS0=0 would typically turn off auto-answer, while ATS0=3 would tell the modem to try to answer after the third ring. (ATS0=1 was surely more frequently used than ATS0=3.) The other way to answer a call is to type ATA (anytime, before or after the incoming call presses in) and then, after the incoming call is recognized (when the modem says “RING”), press the Enter key (to actually send the command) sometime

Wireless communications

Historically, some computers have had cards (plugged into a standard expansion slot) which were designed to communicate with a specific carrier (or type of carrier, such as a provider of “cable” TV services). However, that practice has largely fallen out of use. Conveniently, what people typically do instead is use a standard method of communication, such as using an Ethernet cable or Wi-Fi access, to communicate to a box. Then, the more proprietary/specific type of circuitry is in an external box. This allows compatability with mulitple devices. Some mobile equipment (such as a laptop computer) may have built-in functionality that is used primarily/solely by a single provider in the area. (Such equipment is typically also capable of communicating to an external converter box.)

Mobile phone provider

If a data plan is purchased, phones may often support communications to the Internet. Phones may be able to directly utilize such Internet connectivity, allowing features such as running a web browser on the phone. Also, phones may support an ability to provide other devices with the ability to communicate over that Internet access. This ability is sometimes called “tethering” the Internet access. As phones have started to use standardized (micro-USB) connectors and standardized local wireless communications (Wi-Fi), other devices have frequently been able to use this sort of Internet access. (When Wi-Fi is being used to implement tethering, this has been known to substantially drain the battery of the phone. However, when micro-USB is utilized, the phone might be able to receive electricity to be able to charge its battery. So the phone provides Internet access to what it is connected to, and that device provides electricity to the phone. (If that device is a laptop, it may drain the generally larger battery that is inside the laptop.)

WiMAX / WirelessMAN

Standardized by IEEE the 802.16 family of standards. Wikipedia's article on IEEE 802.16 says:

“Although the 802.16 family of standards is officially called WirelessMAN in IEEE, it has been commercialized under the name “WiMAX” (from "Worldwide Interoperability for Microwave Access") by the WiMAX Forum industry alliance.”

WiMAX has been used by CLEAR, the new name for a company that had been called Clearwire (and, as of 2013, is still called Clearwire in Belgium). IEEE 802.16e-2005 supports speed called “4G”. Wikipedia's article on WiMAX: “Mobile phones” section notes that “Sprint Nextel announced at CES 2012 that it will no longer be offering devices using the WiMAX technology due to financial circumstances”. Sprint's significance in WIMAX is seen by the fact that Sprint owns 50.8% of the company named “CLEAR”.

Infrastructure devices
[#neticons]: Network infrastructure icons

Some icons representing equipment have become so widespread that they are even considered standardized. Cisco icons ( contains some icons that are referenced in the upcoming section. Cisco's web page said “You may use them freely, but you may not alter them.”

See also: local directory/folder for Network icons package(s).

In addition, here are some other icons to mention:

A cloud is typically represented by... a cloud.

A database is represented by a cylinder (similar to a router, but without the arrows). Cisco's package has this in a file called “relational database.jpg”.

Basic connectivity devices
[#netswitc]: Switch

workgroup switch.jpg
from Cisco's
network icons package
(Other icons for “switches”
are included in Cisco's package.)

A switch is, essentially, a splitter (similar in concept to a splitter of electrical outlets). Unlike a hub, a switch contains a processor to do analysis of the traffic so that traffic is only sent out the network cable that is required. The MAC addresses of incoming traffic get stored in a table called the “content addressable memory” (“CAM”) table. The CAM table also stores what connectors (Ethernet port numbers) have been used to communicate with that traffic. The result is that most traffic only needs to be sent out a single port, unlike a hub which typically sends all traffic out all ports (or all ports that the traffic did not arrive on). This permits the switch to become the perimeter of a “collision domain”, which refers to which devices get affected by an Ethernet collision. With a hub, many more devices are typically affected by an Ethernet collision.

A switch typically also has the ability to interpret reflections caused by a collision, and to understand what incoming data caused those collisions. As a result, the collisions are not typically very harmful and the switch is able to support a “full duplex” connection. (That means the switch supports simultaneous two-way communication.) This is also different between a switch and a hub.

In general, more advanced hardware such as a router or even a hardware-based firewall device can effectively perform the basic functionality of a switch. However, there may be a couple of advantages to using a switch:

  • Basic switches are often cheaper than other devices. Many switches may come with more interface ports: many businesses have purchased a 48-port switch. In contrast, some other devices like a firewall will often provide about 4 or 5 ports.
  • More advanced devices (especially hardware-based firewall devices, although this can also be true of routers) may often require some pre-configuration for a specific port. Very often, some ports will have special meanings. In contrast, a managed switch can provide special handling (like 802.1q VLANs) for some ports, but typically the defaults of switches (including managed switches, and most especially including unmanaged switches) is to just allow communication. This typically offers an easier plug-and-play type experience.
  • High-end switches may utilize the concept of an “application-specific IC” (“ASIC”) (“IC” stands for “integrated circuit”: basically a “chip”). Such specialized equipment may help to more quickly get traffic relayed based solely on whatever information is provided by the MAC address (and ignoring the Layer 3 network address, as that would simply be part of the frame's payload). Such relaying may even start before an entire frame is received.
Types of switches

There are some classifications of switches that may be useful to know about:

Dumb/unmanaged switch

A switch that simply relays traffic like all other switches. Such a switch may not even have a MAC address (and, therefore, will not ARP (and has no IP address).

Cisco's icon package (available in the section about network infrastructure icons) contains a standard picture (“workgroup switch.jpg”). It consists of a box that has a couple of arrows moving in both horizontal directions. A rather traditional variation is a 2-D variation that shows a square with such arrows, or perhaps even simply a square. (Unfortunately, a square is quite a simple drawing, and so may often have other purposes. For example, squares may often represent computers. However, when there are squares and circles, squares are more likely to be switches.)

Managed switch

layer 2 remote switch.jpg
from Cisco's
network icons

A managed switch is simply a switch that has a MAC address and so direct interaction with the switch is possible. Generally, the managed switch also has an IP address. The switch may be able to be configured to perform advanced features such as enabling Spanning Tree Protocol, VLANs, and/or mirror/monitoring ports. Another possible/common feature may be and might respond to HTTP.


Cisco's network infrastructure icons package has a special icon for “programmable switch.jpg”. Or, perhaps “layer 2 remote switch.jpg” which may be applicable? (See also: “multilayer remote switch.jpg”?)

[#mltlyrsw]: Multilayer switch
Files from Cisco's network icons package:

layer 3 switch.jpg

multilayer remote switch.jpg

This term refers to a switch that pays attention to more than two layers of the OSI model. (All switches do pay attention to multiple layers of the OSI model: Layer 1 and also Layer 2, so they do fit the description of the English term “multilayer”. However, perhaps simply by convention, the term “multilayer switch” refers to having at least one more layer beyond those two.)

Some switches are referred to as a “multilayer switch”. A multilayer switch may be able to perform some Layer 3 functionality. This sort of terminology is confusing: a device that connects other devices, and which performs layer 3 functionality, may be better off being called a “router”. It seems probable that the difference between a layer 3 switch and a router is more of a marketing term: higher-end equipment using a design specializing in supporting layer 3 routing would be a router, while devices optimized for layer 2 traffic (but capable of communicating with layer 3) might be considered to be a “multilayer switch”.

Further discussion about contrasting a router and a multilayer switch

Although the author of this text has met a small number of technicians (perhaps two) who remember being told that there was an actual technical difference between a layer three switch and a router, at the time of this writing there hasn't been a single one who recalled what that difference was. Slashdot comment posting about layer 3 switches says the difference is simply “what primary task the device has been designed for” and some things in a layer 3 switch are done “differently by default than in a router”. Perhaps any significant difference found “is only a rule of thumb...”, to the point that it “is practically meaningless to distinguish between the "Router" and the "Switch"”. Though there may be some differences in features, it sounds like these differences are not signficant. “In the modular devices, you can pretty much use the modules to configure router as switch and vice-versa, so "L3 Switch" and "Router" have no real difference” between devices that use those terms. To summarize, the author of that comment used quite a few technical terms and so seems to be familiar with some of the different features, yet concluded that any differences are not a really big deal from a technical level. This reinforces the idea that the differences may really be more about marketing than technical ability.

If you really want to just keep things simple, you can follow this guideline: if the thing routes, then it is correct to call it a router, because multiple IETF RFC documents say so, including RFC 4861, and some others mentioned by glossry entry for “router”. Those who wish to complicate things, based on terminology that may be preferred by a hardware vendor, can be instructed that such usage is not standardized across the entire networking industry, and that the definition provided by some IETF RFC documents is simpler to understand and to use.


Cisco's network infrastructure icons package has a special icon for “layer 3 switch.jpg”. There is also a “multilayer remote switch.jpg”.

[#automdix]: Auto-MDIX (MDI and “MDI-X” (“MDIX”))

Some older switches had ports that could only accept a patch cable (common) or a crossover cable (much less common). Some switches (network devices) might even have a toggle switch (such as a button that can be pressed) to alternate between accepting a patch cable or a crossover able.

Towards the end of the time when 100Mbps switches were being sold, a feature called “auto MDIX” came out. This feature simply allowed the switch to detect whether the cabling was a patch cable or a crossover cable. If the untraditional type of cable was used, internal circuitry could be altered so that the communications could work anyway. So, when this feature was used, this technology ended the need to be concerned about whether a cable was a patch cable or a crossover cable. This feature could be found on some later, higher-end 100Mbps equipment.

Common knowledge has stated that the Auto-MDIX is part of the official Gigabit (1Gbps / 1000Mbps) communication standard. Whether or not that is true (apparently it isn't), what is true is that Auto-MDIX has been commonly supported by nearly all equipment that has been advanced enough to support gigabit speeds.

This may be discussed by the section about crossover cables.

[#ethrnhub]: Hub (for Ethernet connections)
Files from Cisco's network icons package:


small hub.jpg

100baset hub.jpg

The term “hub” can refer to a centralized location. There are various types of hubs, including an Ethernet hub, a USB hub, or a “hub” for public transportation (where many buses meet, or a train station). This section of text is about Ethernet hubs.

A hub is similar to a switch, although it does not process individual frames by checking the MAC address. Therefore, a hub is considered a “layer 1” device, because it does not do anything special with any “layer 2” information. (The term “layer” is referring to the OSI Model.)

A hub may also be referred to as a multi-port bridge. The difference between a hub and a bridge is simply that a hub has more than two data connectors.

Hubs tended to be replaced with switches, and so hubs are typically not faster than 10Mbps. For communications that are at least 100Mbps, switches are typically utilized instead.

One possible advantage of using a hub instead of an unmanaged switch is that all the ports are essentially monitoring/mirror ports: traffic goes out each port. This can be more convenient when wanting to network monitoring/sniffing.

Cisco's network infrastructure icons has “100baset hub.jpg” and “small hub.jpg”.

[#netbridg]: (Network) Bridge

from Cisco's
network icons

A bridge, simply, is a device that performs the action known as bridging. The term “bridging” refers to relaying traffic from network to another network, resulting in a single larger network. This effectively ends up accomplishing a similar task to network traffic routing with one key difference to distinguish the proper use of terminology: “routing” typically refers to traffic being joined by a Layer 3 router, while the term “bridging” refers to connectity layers below layer 3 (operating at layer 2, or perhaps even just layer 1) of the OSI network model. This basically means that the communications are relayed looking at the MAC address (which get transmitted in Ethernet frames), rather than Layer 1.

A standardized icon for a bridge may be look like a box, or a square, but the top is curved higher on the sides than in the middle. The result may look like a non-taunt rope bridge at the top of this shape. With Cisco's package, this may be represented by the bridge.jpg file.

Difference between hub/bridge/switch

The difference between a hub and a bridge may be a bit hazy/vague: different technicians may provide different answers. Likewise, the difference between a hub and a switch might also be a bit less than entirely clear. Although, the difference between a hub (which forwards traffic out all ports) and a switch (which has a chip to intelligently try to decide which port to send traffic out) is clear. A bridge is likely rather in between those two devices.

Some people may say that the term “bridge” may refer to a device that has just two network ports. (e.g. Wiki at post about switch/bridge states, “bridge has only TWO ports.”) This would basically the same in concept as a repeater or a coupler. So, according to this definition, a bridge is basically a two-port hub, and a hub is basically a multi-port bridge that has more than just two ports. However, this is not universally agreed. Later on, the same Wiki at post about switch/bridge article states, “the IEEE specs define bridges... and those bridges have tons of ports, not just two.” forum post by “'nother”, also states, “really, it is *not* correct” saying “Switches are really just multi-port bridges”... “the guts are quite different.”

Another distinction that is sometimes made is that a bridge connects different types of communications, such as Ethernet or Token Ring (even though both Ethernet and Token Ring can communicate with similar hardware). Or, some people may say the opposite.

A bridge and a switch are different. A switch is capable of sending traffic to only a single network port, because it uses a chip (generally an “application-specific integrated circuit”, a.k.a. “ASIC”) to process the traffic. Commonly there are multiple switch ports, so that when there are multiple devices to be plugged into a switch, then each device will be plugged into its own interface. In contrast, a switch might often be plugged into hubs, so the bridge doesn't tend to forward traffic to only a single device. A bridge may be able to notice when traffic from one device is meant for another device on the same side of the bridge, in which case the bridge may be able to not forward the traffic to the other side (according to SlareX forum post). This may effectively reduce the size of collision domains. The primary difference between a bridge and a switch is the presense of the chip in the switch which makes the switch more intelligent. The difference may largely be that the chip in a switch provides superior performance: Some people may consider this distinction to be driven more by marketing (e.g. Wiki at post about switch/bridge states, “The difference bridge/switch is mainly marketing.”) rather than any clear functional difference between those types of devices. For instance, when discussing Spanning Tree Protocol, the roles of the devices will use the term “bridge” (even though people are typically using switches today).

There is a reason why the difference between these similar terms is often not clearly defined. Simply know that a bridge is rather similar in concept to a switch, although may be older technology. Anything further is less clearly agreed upon. IceStorm's forum post states, “This stuff gets really sticky out in the field, and a lot of it is vendor-specific. Generalities don't get you very far when things start crashing down around you.” So, to summarize, the precise definitions are not widely understood or universally agreed upon.

[#wap]: Wireless Access Point (“WAP”)
Files from Cisco's network icons package:

wireless router.jpg


wireless bridge.jpg

A device that handles wireless traffic. Typically a WAP contains an antenna (and possibly multiple), and also acts as a router or a bridge/switch, connecting wireless traffic to a wired network. A WAP may also provide some other functionality, such as firewalling.


Cisco's network icons package contains “accesspoint.jpg”, which basically looks liek a switch but has a couple of overlapped sine waves, and is missing the arrows. (The sine waves are phase shifted, looking a bit like spiraled DNA. Also, the sine wave icon in Cisco's package is on the front of the box, rather than the top.) Other icons include “wireless router.jpg” and “wireless bridge.jpg”. Other files with names starting with “wi” or “wl” or “ant” may be applicable to more specific types of wireless equipment.

Those may be the most promising. There are others, like “mesh ap.jpg”.


A repeater simply takes the signal from one wire, and duplicates it on another wire. This operates on the principles of electrical power, and does not have any need to try to interpret any the data (such as a MAC address or Ethernet jamming) that might be getting communicated with the electrical signal.

A repeater does require power, as it re-amplifies any electrical signal to make up for any electric loss caused by sending a signal through wiring. (If a device does not add any electricity to the mix, it would not be classified as a repeater but rather as a coupler.)


A couple of connectors, typically referring to a device with multiple jacks that wires may plug into. This can effectively be nothing more than a piece of plastic with a bit of copper. This does not require any electricity beyond what is provided by the wiring.

Traffic handling equipment
[#router]: Router

from Cisco's
network icons package

Basically like a switch, a “router” has features related to routing network traffic.

A definition for the term “router” is provided by RFC 4861 section 2.1 which says a router is “a node that forwards IP packets not explicitly addressed to itself.” (In simple terms, the word “node” simply refers to a device that is communicating on the network.)

A router typically reduces the TTL hop count of Layer 3 traffic. (See: Glossary entry for TTL, Glossary entry for “hop”.)

Also, a router typically does not relay broadcast traffic. Therefore, a router ends up being a perimeter device that forms an outer boundary for a “broadcast domain.

Common icons

Cisco's network icons contains router.jpg. It is like a database icon (a cylinder), although it has some arrows at the top. Two of the arrows point towards the center (and so are pointing towards each other), while two other arrows point away from the center. They are also arranged in an X shape. A simplified drawing that represents “router” is either a cylinder with an X on it, or even just a circle with an X in it.

For professional work, the icon ideally has arrows which look similar to a plus sign rotated 45 degrees, more commonly referred to as an “X” shape. Or, if drawing the arrows seems like too much work, at the very least have an “X” on the device. The reason that “X” shape is so important is that a cylinder without that “X” shape has another standardized meaning, and so some people many be naturally inclined to think that it may represent a disk, or some other source of data (like a volume that is stored on part of a disk, or a volume that has data stored on multiple disks). For example, Cisco's package may have a “relational database” icon that contains the cylinder without markings on the top.

Cisco's package also has several variations for specific types of routers. For example, a firewall that also acts as a router may use the “router_firewall.jpg” file.

[#hwfirewl]: Firewall (hardware)

A firewall is a device designed to refuse to forward traffic that is deemed undesired (probably due to a high potential of being harmful). The act of firewalling performs this same functionality, and can often be effectively performed by a router (particularly when it implements NAT).

The section on firewall implementations starts with software firewalls, but then discusses hardware that is specifically designed/marketed to perform firewall functionality.

As for common icons, a firewall may often be rendered as flame, or as flame ontop of a device (like a router). Cisco often likes to use “firewall.jpg” which basically looks like a wall made up of bricks, although that icon (or a shield) may be more appropriate to represent any sort of protection. (This picture of the brick wall may typically be rendered in red.) Another variation is to cover a picture with the texture of bricks, such as the round outside of the router.jpg file (possibly with flame above that). Cisco's package has a variation called “router_firewall.jpg” where it looks like the bricks are on the top of the router (but below the arrows). Also, Cisco's package contained graphics named “pix firewall.jpg” and “cisco asa 5500.jpg”. The “pix firewall.jpg” is also known as “pix firewall right”; there is also a “pix firewall left” as shown at Slideshow of Cisco network icons.

[#netcloud]: a network cloud

from Cisco's
network icons

A cloud clearly is clearly defined to represent a part of a network where specific details may be more hazy, unclear, murky, or fuzzy. This simplifies diagrams, making them more understandable.

Take, for example, two multi-floor buildings located in different states/provinces/nations. A network diagram may show some of the equipment on one of the floors of the building: workstations connect to wall ports that get connected to a patch panel, which then connects to a switch which is then connected to a router which is then connected to a firewall which then passes traffic onto the broadband Internet connection's modem.

Then what happens to the traffic?

Well, an average subscriber doesn't know. All that the customer really knows is that data is given to the Internet provider, and then the traffic ends up appearing at the other building. The Internet provider takes care of the details on how the traffic is delivered. The customer probably does not know how many switches are involved in getting the traffic relayed. The customer might try to figure out how many routers are involved (by using Traceroute). However, even if the customer figured that out, the route may change. The Internet Service Provider may rely on a telephone company, and that telephone company might change some agreement it has with another company that helps to deliver traffic.

The customer does not need to try to document the exact route. In fact, the customer might not be authorized to know that much information about the internal systems of some of the other companies that may be involved. However, that is okay, because the customer does not need to care exactly how the traffic is delivered. The precise details of how the traffic gets relayed may change without the customer's permission or knowledge. That is entirely okay.

Instead of trying to document every piece of equipment, the network diagram may simply specify that information enters a “cloud”. The network diagram doesn't need to document all of the details. Instead, the network diagram may just document the minimum information that the customer needs to know: if information is sent out to one end of the cloud, then that information arrives on the other end of the cloud. As long as the customer keeps paying for Internet service, and the cloud functions as expected (by delivering the information where it needs to go), that may well be “good enough”.

In fact, a company which owns a multi-floor building might even use a “cloud” for part of its internal network. The detailed network diagram for the third floor might specify that information gets sent to a router, and then to a “cloud”. That cloud may be given a name, such as “secondary subnet system number two”. Then, if people do need further information, they may check some separate documentation for “secondary subnet system number two”. By using this approach, if changes do get made to “secondary subnet system number two”, then the only documentation needing any updates will be the documentation for that one specific network. All the other network diagrams, which refer to the system simply as a cloud, will still be up to date and accurate, and not need any changes. (Trying to document everything in each floor's network diagram, could require that multiple pieces of documentation need to be updated when a change is made.

The term is widely used. Network providers may often refer to a “cloud” of another company, which simply refers to a bunch of equipment that is handled by that other company. A company in Northwest America that provides Fiber-optic network connectivity has called itself “FiberCloud”.

Cloud service has sold books, and other items often found in bookstores (such as music), and now sells many other items (such as computers). They are a major retailer on the Internet, and have built up a substantial infrastructure.

At some time, decided to basically start renting out usage of some of its substantial computer and network infrastructure. Customers started paying to get access to virtual machines. People wanted to have their data backed up. This could often require more bandwidth if people started trying to copy their virtual machine's data to other computers that were not part of's network. let people know that their data was being backed up by

People started asking questions, like “How many physical computers are being used?” Another question may be, “How is the data being backed up?”

The answer, basically, was “Don't worry about it.” The idea is that subscribers are paying for a service. It is up to to worry about the details to be able to reliably provide the service. Similar to how an Internet service customer doesn't need to know how the Internet traffic is delivered: all that matters is that it is successfully, reliably delivered.

This concept was so similar to the concept of clouds on a network diagram this service started becoming known as “cloud” service. named their service “Elastic Compute Cloud” (and often abbreviated it as “EC2”). ( has since renamed the platform to “Amazon Web Services”.)

Other companies have hopped on board as well. Wikipedia's article on “Amazon Elastic Computer Cloud”: “See also” section and Wikipedia's article for “Google App Engine”: section labelled “Competition” and Wikipedia's article for Azure: section called “Competitors” list several options, including some with the word “cloud” as part of the company name or service name: Rackspace Cloud, Enlight cloud, HP Cloud Services, Lunacloud (headquarted in London).

Since then, software such as Eucalyptus and OpenStack provide software solutions for people to be able to run a “private cloud”. Naturally, many organizations who use a privately-run “cloud” will want solid documentation about their equipment being used. The term “cloud” seems to have migrated a bit from its earlier meaning of being imprecisely defined, and now refers to the use of virtual equipment that can be created fairly easily without much concern about having a physical machine for each piece of virtual equipment.

Specific implementations

One company is noted for having a lot of equipment installed in high-end locations, like (seemingly famously) equipment that runs in the New York Stock Exchange. This company is named Cisco.

Cisco equipment often uses a custom operating system that is provided by Cisco. More information may be found from: Cisco professional-level equipment and/or other resources such as: Cisco Intro and/or standard warning about using Cisco equipment and/or basic IOS usage guide.

[#netcable]: Cabling

See: Network cabling/wiring

A lot of information about the topic of “network cabling/wiring” was here before that topic was moved to its own sub-section. Here are some hyperlink anchors that were previously shared with the public, and now direct people to the newer, more specific location. (This list also shows some of the topics that are discussed in the section on network cabling/wiring.)


This is discussed in another section about network connectors.