Networking Fundamentals
Performance Metrics
Bandwidth:
A measurement for the hypothetical rate at which data could travel over a network as dictated by the transmission medium being engaged (e.g. fiber, cellular, or DSL).
Unit of Measure: bits per second, commonly reflected in kilobits per second (Kbps) or (Mbps).
Latency:
Any sort of delay (lag) that occurs during the transmission of data. All transfers have lag because of the inherent constraints of a signal traversing its physical medium over any given distance.
Lag is introduced by other factors as well, including the number of connected users and the complexity of requests being processed over the network.
Unit of Measure: seconds, usually milliseconds (ms).
Throughput:
The actual observed rate at which data travels through a network over a set amount of time. This is limited by both latency and bandwidth as well as a variety of contextual variables.
Throughput can be thought of as the observable, real-world rate of data transmission (bandwidth) determined with network overhead taken into consideration.
Unit of Measure: bits per second, commonly reflected in kilobits per second (Kbps) or (Mbps).
IP Addresses:
As the specific address used to locate a device on a network, an Internet Protocol (IP) address is a unique identifier much like a mailing address. There are many technological standards currently regulating IP addresses that may be relevant depending on the context.
The most prevalent syntax for IP addresses is currently IPv4. Devices are given a numerical IP address when they connect to a network, thereby "mapping" each device as a reachable endpoint on the Internet or local network. An IPv4 address is formated as #.#.#.# (where # can be anywhere from 0 to 255).
It is important to note that there is a limited number of IPv4 addresses -- due to the fixed mathematical constraints of listing a number, "#," from 0 to 255 four times sequentially -- nearly 4.3 billion unique, assignable IPs in total. The rate of depletion for IPv4 addresses correlates directly to amount of Internet-enabled devices seeking connections.
IPv6 alleviates IPv4 address exhaustion. It is a 128-bit address format (whereas IPv4 was 32-bit), meaning the maximum number of IPv6 addresses is over 7.9 x 1028 times greater than the number of addresses available using IPv4. Its syntax is alphanumeric (technically hexadecimal) -- formatted as eight groups of four characters -- 2001:0db8:85a3:0000:0000:8a2e:0370:7334 for example.
IP addresses are assigned differently when a device is connected to a LAN as opposed to a WAN. This difference in setting (LAN vs WAN) establishes whether an IP address can be considered private or public.
Private IPs are strictly distributed to devices on a LAN. Since local networks are only reachable by devices with a direct connection to that network, these private addresses only need to be unique within the LAN itself. This means that all LANs may leverage the same private IPs (internally). It is with this in mind that the following range of IPv4 addresses have been reserved for private/ LAN use:
| RFC1918 Name | IP Address Range | Number of Addresses |
|---|---|---|
| 24-bit block | 10.0.0.0 – 10.255.255.255 | 16,777,216 |
| 20-bit block | 172.16.0.0 – 172.31.255.255 | 1,048,576 |
| 16-bit block | 192.168.0.0 – 192.168.255.255 | 65,536 |
If you see an IP address that falls within the ranges listed above, it is the private address of a device connected to a LAN.
Public IPs are used to assign WAN endpoints. Unlike private addresses, they are globally routable and used for communication between hosts on the Internet and WAN gateways that feed into the various LANs distributed across the world (such as the router you're likely connected to right now). Any address that isn't assigned for private use is considered public.
A Static IP is reserved by the network for a particular device. This means that the device with a static IP will be reachable by the same address every time. The opposite of a static IP is a Dynamic IP, which implies IP addresses are being assigned to devices based off of availability (i.e. if a laptop is shut down, the machine's IP will be released and reassigned to another device that joins the network, and the laptop itself will likely have a new IP once it comes online again). This can be problematic for network engineers because they often rely upon static routes that need to deal with fixed variables.
Only the ISP can supply a static public IP, and there is typically a surcharge (cost varies provider to provider).
Data must be routed through a public IP for it to flow between LAN and WAN environments. Network Address Translation (NAT) is the method through which private (LAN) IPs are mapped to a public-facing (WAN) address so traffic can pass to and from the local network and the Internet.
This technique helps combat the scarcity of IPv4 addresses through one-to-many NATing, where one Internet-routable (public) IP address is mapped to multiple LAN endpoints -- NAT makes it so private networks don't need a public IP for every LAN-connected device that requires Internet access.
The opposite of NAT, where a connected device is assigned the IP directly obtained from the ISP. A device connected via IP Passthrough forfeits LAN access for direct WAN connectivity.
Common Network Appliances:
Routers:
Routers handle network traffic as well as the assignment of IP addresses in most common configurations. Devices connected to a router may receive a private IP address and are all considered part of the same LAN; devices can connect over Ethernet or WiFi.
Modems:
Occasionally supplied by ISPs, a modem provides Internet access by transmitting data to and from a connected device via the modem's interface. Traditional (wireline) modems use a physical connection as the interface medium -- a coaxial or fiber-optic cable in most cases -- while cellular (wireless) modems rely upon radio frequency (RF) signal.
Standalone modems are not commonplace since most ISPs supply routers with built-in broadband interfaces. The Internet access supplied by a modem can give WAN access directly to a single device, or multiple devices when fed through a router.
WAN Failover is accomplished by having a backup Internet connection ready to kick in should the primary means of WAN access go offline. A second modem is required to facilitate this.
NOTE: Whether built into a second router appliance or existing as a standalone device, two modems are needed.
Switches:
A switch is a network appliance that houses multiple Ethernet ports but does not handle any sort of routing. These devices are used to expand the number of ports available to a router without requiring a replacement router to acquire more ports (since switches are less expensive).
Firewalls:
Hardware firewalls are purpose-built appliances designed to provide network security. While some degree of control and protection can be achieved using the features included in a traditional router, advanced applications will begin to strain the resources of said router. A dedicated firewall can be installed to gain additional security precautions.
MAC Addresses:
A media access control (MAC) address is a unique identifier associated with a device's network interface; they are assigned by the hardware manufacturer and are static by nature. As equipment joins a network, the IP address obtained from the ISP (or LAN router) becomes associated with a MAC address belonging to each connected device.
All network-enabled electronics -- computers, cell phones, printers, firewalls, etc. -- have at least one MAC address, including Accelerated LTE Routers.
Firmware:
Hardware consists of the physical components belonging to a digital system. Software is the set of instructions that spell out how that system can function.
Firmware, in turn, is the underlying logic that ensures each hardware component is capable of properly fulfilling the instructions provided by software.
Data Center:
A centralized location for network infrastructure. At smaller locations, this could be a telecommunications ("telco") closet or an on-site server room. Larger enterprises have off-site data centers that are managed remotely. A hardened data center is designed to stay online in the face of catastrophic events.
As more and more services migrate "to the cloud" (by way of Office 365, OneDrive, etc.) businesses are moving away from on-site data centers, and even letting go of internally managed remote data centers, for subscription-based Infrastructure-as-a-Service.