Computer Network Card

Сетевая карта компьютера

Let's start our topic with the fact that network cards are different and can differ, both in the spectrum of solved problems and in form-factor (appearance). A network card is also often called a Ethernet (Ethernet) controller, a network or NIC (Network Interface Card) adapter.

First of all, let's divide the network cards into two large groups:

  • External network cards
  • Built-in or integrated (onboard)

Let's start with the outside. From the name itself it follows that network cards of this type are installed in the system unit of the computer additionally (a separate expansion card) or as another external device.

First, let's talk about PCI network cards. The abbreviation stands for Peripheral Component Interconnect or I / O bus for connecting peripherals to the computer's motherboard . These cards are named for the reason that they are installed in one of the PCI slots (connectors). Here in these, in fact:

The PCI interface itself has a peak bandwidth for a 32-bit version running at 33.33 MHz at 133 MB / s, a 3.3 or 5V connector voltage. It is used to install additional expansion cards (old video cards, modems, network adapters, TV tuners of various video capture cards and video conversion, etc.) to the computer.

So, what kind of network cards are installed there? But these are the most usual dollars for five or six:

There are adapters and other varieties - Wi-Fi (for organizing wireless networks).

As you can see, the connection interface is one (PCI), and the principle of operation is different.

Now, due to the gradual "withering away" of this interface, network cards of the form factor "Pci Express 1X" are produced.

PCI Express 1x Network Card PCI Express 1x Network Card

This is the case with external network cards. There are also built-in (integrated in the motherboard) cards. Check the presence of the built-in network by looking at the back of the system unit .

Built-in network card Built-in network card

Here we can visually observe the output of an integrated network card. Next to the connector for a twisted pair set one or more information LEDs, which can be oriented on the presence of connectivity and overall network activity.

By the way, these LEDs can indirectly form an idea of ​​the device's performance. I will explain my idea: when the computer is turned on and the network cable (twisted pair) is connected to the card, the LED on it flashes, as it is called, in time with the reception (transfer) of data packets by the adapter into the network.

With a non-working network adapter, the behavior of the indicators can be as follows:

  1. Not one of the LEDs glows at all
  2. LED constantly "lit" (not blinking)
  3. The indicator flashes, but it is absolutely monotonous. The period and amplitude of this "blinking" are the same throughout the time

So - note these moments. In details - our everything! :)

If you go further and open the computer case , then inside (not far from the connector) you can find a chip that is soldered on the motherboard, which implements all the functions of the network adapter.

By marking, we see that this is a RTL chip (from Realtek) under the number 8211BL.

Note: built-in solutions, unfortunately, do not differ reliability. In our organization, for example, outages of integrated network cards happen regularly. I can not say that often, but - it is stable. By the way, my (purchased half a year ago) working computer just a few days ago burned the network card, which once again reinforced my opinion about the unreliability of integrated components. I had to put the outside.

I want you to carefully take a closer look at the following photo:

Here we look inside the connector of the network card. Do you notice the difference? In one connector (in the photo - on the right) there are four contact pads, and in the other (on the left) - eight. And both cards are designed for the speed of transmission over the network of 100 megabits per second.

What is the catch? And he, on any, here is present :) Let's remember with you what the twisted pair cable looks like, see the photo.

On the right, it's called a UTP cable (Unshielded Twisted Pair - unshielded twisted pair). The fact that it is twisted (twisted), we can clearly see from the photo above. Some of its conductors are entwined with each other to improve the noise immunity of the entire cable as a whole.

The designation "unshielded" implies that there is no additional protective shield (braid) made of foil or metal over the veins. Again - for better cable security. A "pair" because the conductors in the cable are twisted in pairs and - in colors (white-orange - orange, white-green - green, white-brown - brown, white-blue - blue).

Now - the most important thing: to ensure data transfer over the network at a speed of 100 megabits per second, you do not need to use all four pairs (eight conductors-veins), enough two pairs (four cores)! And strictly defined numbers are used: the first, second, third and sixth postings.

Directly from the side of the RJ-45 connector, it looks like this:

Crimping of twisted pair UTP RJ 45 Crimping of twisted pair UTP RJ 45

According to the above, to provide a speed of 100 megabits, we use "veins" under the numbers 1, 2, 3 and 6. We look at the figure above. It's two pairs: orange and green.

Note: Of course, it is up to us to decide which veins to use when fixing the cable. The main thing to remember is that this should be the 1st, 2nd, 3rd and 6th conductor (for networks with a transmission speed of 100 megabits / s).

And now again look at the photo, which shows the connectors of network cards of the computer. On the right image there are only four contact areas: the first, the second, the third, the next two - are skipped and further ... what? Correctly - the sixth! :)

When are all eight sites used? In networks with a transfer rate of one gigabit per second (and higher). That's where all the conductors of the network cable are used to the fullest :)

So, something we with you (or rather I am one :)) "urolil" away from the main topic. What else are network cards? Let's look at an external adapter for a laptop based on the PCMCIA standard. This is an external expansion board that plugs into the corresponding slot.

It stands for "PCMCIA" as a Personal Computer Memory Card International Association (international association of computer memory cards). Initially, the standard was developed for memory expansion cards. After some time, the specification was expanded and it became possible to use "PCMCIA" to connect various peripheral devices. As a rule, it connects network cards, modems or hard disks through it.

Imagine an unpleasant picture: your laptop (three times "fuck" it to the left) has a built-in card in it. What to do? The solution is in the photo below:

PCMCIA adapter for laptop PCMCIA adapter for laptop

There are, however, other solutions that are suitable not only for mobile computers, but also for stationary ones. This is USB network cards.

They can be performed in different ways, but the principle of their work does not change from this. Here, for example, two such devices in the photo below:

USB Network Adapter USB Network Adapter

Or even like that, it's more like a flash drive :)

USB Network Adapter USB Network Adapter

On this was going to finish the article, but ... changed his mind! :) I also wanted to tell you about such a variety of external network cards, as server network cards, which are used in high-performance systems and have more advanced (in comparison with conventional adapters) networking capabilities.

Typically, they have a standard connection interface - PCI (or its extended version - PCI-X). Here, for example, server network card "D-Link DFE-580TX".

Server network card photo 1 Server network card photo 1

As you can see, this, in fact - four network adapters, combined in one physical device. Each of the four network ports (cards) has its own MAC address (a unique 12-digit physical identifier for any card or other network device). At the same time, one logical identifier (IP address) can be assigned to the entire port group. For an operating system, a group of such cards looks like one virtual card.

Note: MAC (Media Access Control) address is also often referred to as a physical or hardware address (Hardware Address). For example: The MAC address of my network adapter at work is 00-2C-13-C3-A3-65. The network can not have two identical hardware addresses. You can find it on the command line: ipconfig / all or by such a wonderful command using the same name utility, like getmac. Getmac in very convenient and visible form will show you all MAC addresses of all network devices installed in the computer.

Let's continue. Combining several cards into one becomes possible when using the "Port Aggregation" technology (aggregation or - port aggregation). Combining ports means merging several segments of the network into one that has more performance. When several network ports form one virtual port, then its throughput (theoretically) is equal to the performance of an individual port, multiplied by their number.

Server network cards can work in two main modes. Let's look at them in more detail. Using the software bundled with cards of this class, you can configure each port as "active" (load balancing mode) or reserve any ports to provide fault tolerance (recovery mode).

The load sharing mode distributes the network traffic evenly across the active segments, reducing the overall load on the adapter, and the recovery mode (in the case of a physical disconnection) ensures uninterrupted communication between the network card and the network.

What is still good server computer network card? Depending on its "tightness" :), it can implement computational functions (counting and generating checksums of data frames transmitted over the network) in hardware, without adding additional CPU to the computer .

These adapters are equipped with specialized LSI (Large Integral Circuits), which take a significant part of the work (collision detection, assembly, disassembly of data packets, verification of frame checksums and retransmission of corrupted packets). Thus, as we have already said, a significant part of the load is removed from the processor, which in the server system already has something to do :)

Moreover, expensive server network cards install their own processor. Such cards show very good performance, because they can effectively handle even with a heavy load. The presence of its own processor allows you to install on them up to one megabyte of RAM . And this already translates these products from the category of simply network cards to the category of communication network processors.

Also it is necessary to note such useful function, as self-repairing drivers of similar devices. What it is? For example, after a network failure, the adapter can decide whether to restart the network card driver, enable network connection integrity check, or even forcefully disconnect the failed port.