What is a network, again? Network engineers are just like road builders. We've been building these roads since the start of time, and through new technologies we build bigger and faster roads. As a network engineer, our goal is to make these roads as fast and efficient as possible.
Hardware can be made by anyone. Companies can make really fast routers and other hardware, but what makes Cisco stand out is their operating system; the IOS (Internetwork Operating System). Companies are now making their software appear and behave more like IOS, which makes Cisco certification more and more valuable.
Let's now talk about what makes a basic everyday network. The common hardware components involved in modern networks are:
1. Switch - The switch provides local area connectivity. All devices plugged into a switch reside in the same LAN. The LAN provides the most basic form of network communication, allowing sharing of resources (think of resources as data, servers, printers, and so on).
3. WAP - Wireless Access Points allows wireless connectivity (through the 802.11 series of standards). The biggest challenge of WAP in the past was security, but modern security protocols are now secure enough to be implemented in enterprise settings. However, wired access is still the preferred mode of communication for its speed, reliability and relative security.
4. Clients and Servers - These are the actual users of the network.
Network equipment are usually of standard sizes called Rack Units. Racks and equipment have RU ratings. A 24 RU rack can hold 8 switches of 3 RU. Racks typically have attached patch panels, which aid in cable management and organization.
The most common type of LAN is the Ethernet LAN. The most common Ethernet LAN cable is Cat 5e. These cables terminate with an RJ-45 connector. Cat 5e cables consists of 4 twisted pairs, and if crimped properly in the correct order, create an RF insulation field which allows maximum speed connectivity up to 100m. The common cable wirings are: Straight, Crossover and Rollover. The two standards for wiring are T568A and T568B.
Fiber is now more prominent in our everyday networks with the introduction of OpenNET. Fibers make use of light to communicate, which are much less prone to interference. These cables come in Single- and Multi-Mode types. The Single-Mode fibers are glass fibers which allows transmission of light across miles before requiring regeneration/repetition. The bandwidth allowed on each fiber cable is extremely high due to the number of ways we can exploit light, and newer light modulation techniques are being constantly designed.
In some older commercial networks we could encounter serial connections. Routers are connected to a CSU/DSU connected to wall ports, which is akin to a modem for serial connections. These are slowly being replaced by Ethernet fiber.
When we talk about data speeds in network, we need to know the units of data size. The smallest unit of representation is a bit, which has two states (0 or 1). When you put 8 bits together, you get a byte, which is equivalent to an ASCII character. A kilobyte is a 1024 bytes. A megabyte is 1024 kilobytes. Following the megabyte is the gigabyte, terabyte, petabyte, exabyte, zettabyte, yottabyte, brontobyte and gyopbyte. The world had 160 exabyte of data in 2006, 500 exabyte in 2009, and in 2012 the world was estimated to have 2.7 zettabytes of data.
When we look at data size, we usually talk in bytes. However, when referring to data rates, we usually talk in bits. A 100Mbps connection is actually a 12.5MBps connection. In simple mathematics, assuming 10% overhead (files are typically chopped into packets of 1500 Bytes, which is encapsulated with header overheads), to send a 100MB file over the network, we'll take...
100MB/(100*0.9/8) = Approximately 8.9 seconds.
When we go into networking, we need to be able to abstract the functions of devices into different groups. The standard architecture to describe network communications is the OSI Model. It is a standard to create standards. This abstraction allows interoperability, and implementation of compatible protocols.
OSI used to be a protocol, instead of just a model. It was developed in 1977 to compete with the TCP/IP stack. OSI was much better, but TCP/IP was chosen as the industry standard because of the difficult address scheme. The OSI's addressing scheme is actually similar to IPv6, which was created because IPv4 address space is running out.
The OSI model has 7 layers:
7. Application
The application layer is where the simplest form of data is created. API's reside on the application layer to handle the top 3 layers, which allows the programmers to focus on his program and less on the way the actual mechanics involved in sending data.
6. Presentation
Formatting and cryptography is part of the presentation layer. It defines how to format data (such as a picture, video, sound, or other data) into standard and generic format recognizable by the receiving end.
5. Session
This handles the entire session of access, which may comprise multiple connections.
4. Transport
The two most prominent transport protocols are the TCP (connection-oriented and reliable) and UDP (connectionless and unreliable). TCP is used when the integrity of data is important, while UDP is used for realtime applications where time is sensitive. This allows application/service separation through ports.
3. Network
It provides logical addressing and routing. Path determination is performed in the network layer which allows traversing of packets across networks.
2. Data Link
It is required for local communication in LANs and devices in the same subnet.
1. Physical
This describes the actual signals and hardware required to transmit these data. These include the hardware boxes itself, the cables or wireless signals, etc.
For data to be sent, it is encapsulated, which means it moves down the OSI Layer starting from Application. The application first creates the request, which is passed to be formatted by the presentation layer. The session is then established in the local computer. The transport layer then creates a logical connection through a transport protocol (e.g. TCP) which opens a source port (Registered/Dynamic 1024 to 65535) on the local system destined for a known open destination port of the target system (e.g. Port 80 if you're accessing a web server, and port 443 for secure connections). The data is then chopped into small TCP segments with TCP headers, then encapsulated with an IP header (with source and destination IP and other information) in the network layer to become a packet. Finally, the data encapsulated with an Ethernet header (source and destination MAC addresses and other information) to become a frame. At Layer 1 it is transmitted into the network.
No comments :
Post a Comment