Hardware component that connects a computer to a network
"Network card" redirects here. For the British Rail discount card, seeNetwork Railcard.
Network interface controller
A 1990sEthernet network interface card that connects to the motherboard via the now-obsoleteISA bus. This combination card features both aBNC connector (left) for use in (now obsolete)10BASE2 networks and an8P8C connector (right) for use in10BASE-T networks.
Anetwork interface controller (NIC, also known as anetwork interface card,[3]network adapter,LAN adapter andphysical network interface[4]) is acomputer hardware component that connects acomputer to acomputer network.[5]
Early network interface controllers were commonly implemented onexpansion cards that plugged into acomputer bus. The low cost and ubiquity of theEthernet standard means that most newer computers have a network interface controller built into themotherboard or contained in aUSB-connecteddongle, although network cards remain available.
Modern network interface controllers offer advanced features such asinterrupt andDMA interfaces to the host processors, support for multiple receive and transmit queues, partitioning into multiple logical interfaces, and on-controller network traffic processing such as theTCP offload engine.
The network controller implements the electronic circuitry required to communicate using a specificphysical layer anddata link layer standard such asEthernet orWi-Fi.[a] This provides a base for a full networkprotocol stack, allowing communication among computers on the samelocal area network (LAN) and large-scale network communications through routable protocols, such asInternet Protocol (IP).
The NIC allows computers to communicate over a computer network, either by using cables or wirelessly. The NIC is both a physical layer and data link layer device, as it provides physical access to a networking medium and, forIEEE 802 and similar networks, provides a low-level addressing system through the use ofMAC addresses that are uniquely assigned to network interfaces.
12 early ISA 8 bit and 16 bit PC network cards. The lower right-most card is an early wireless network card, and the central card with partial beige plastic cover is a PSTNmodem.Intel Ophir 82571 dual-port Gigabit Ethernet controllerdie
Network controllers were originally implemented as expansion cards that plugged into a computer bus. The low cost and ubiquity of the Ethernet standard means that most new computers have a network interface controller built into the motherboard, and that method is called LAN on motherboard (LOM). Newerserver motherboards may have multiple network interfaces built in. The Ethernet capabilities are eitherintegrated into the motherboardchipset or implemented via a low-cost dedicated Ethernet chip. A separate network card is typically no longer required unless additional independent network connections are needed or some non-Ethernet type of network is used. A general trend in computer hardware is towardsintegrating the various components of systems on a chip, and this is also applied to network interface cards.
An Ethernet network controller typically has an8P8C socket where the network cable is connected. Older NICs also suppliedBNC, orAUI connections. Ethernet network controllers typically support10 Mbit/s Ethernet,100 Mbit/s Ethernet, and1000 Mbit/s Ethernet varieties. Such controllers are designated as10/100/1000, meaning that they can support data rates of 10, 100 or1000 Mbit/s.10 Gigabit Ethernet NICs are also available, and, as of November 2014[update], are beginning to be available oncomputer motherboards.[6][7]
Modular designs likeSFP and SFP+ are highly popular, especially forfiber-optic communication. These define a standard receptacle for media-dependent transceivers, so users can easily adapt the network interface to their needs.
LEDs adjacent to or integrated into the network connector inform the user of whether the network is connected, and when data activity occurs.
Direct memory access (DMA), where a device other than the CPU assumes control of thesystem bus to move data to or from the NIC to memory. This removes load from the CPU but requires more logic on the card. In addition, a packet buffer on the NIC may not be required, andlatency can be reduced.
Multiqueue NICs provide multiple transmit and receivequeues, allowing packets received by the NIC to be assigned to one of its receive queues. The NIC may distribute incoming traffic between the receive queues using ahash function. Each receive queue is assigned to a separateinterrupt; by routing each of those interrupts to differentCPUs orCPU cores, processing of the interrupt requests triggered by the network traffic received by a single NIC can be distributed, improving performance.[10][11]
The hardware-based distribution of the interrupts, described above, is referred to asreceive-side scaling (RSS).[12]: 82 Purely software implementations also exist, such as thereceive packet steering (RPS),receive flow steering (RFS),[10] andIntelFlow Director.[12]: 98, 99 [13][14][15] Further performance improvements can be achieved by routing the interrupt requests to the CPUs or cores executing the applications that are the ultimate destinations fornetwork packets that generated the interrupts. This technique improveslocality of reference and results in higher overall performance, reduced latency and better hardware utilization because of the higher utilization ofCPU caches and fewer requiredcontext switches.
With multi-queue NICs, additional performance improvements can be achieved by distributing outgoing traffic among different transmit queues. By assigning different transmit queues to different CPUs or CPU cores, internal operating system contentions can be avoided. This approach is usually referred to astransmit packet steering (XPS).[10]
Some products featureNIC partitioning (NPAR, also known asport partitioning) that usesSR-IOV virtualization to divide a single 10 Gigabit Ethernet NIC into multiple discrete virtual NICs with dedicated bandwidth, which are presented to the firmware and operating system as separatePCI device functions.[3][16]
Some NICs provide aTCP offload engine to offload processing of the entireTCP/IP stack to the network controller. It is primarily used with high-speed network interfaces, such as Gigabit Ethernet and 10 Gigabit Ethernet, for which the processing overhead of the network stack becomes significant.[17]
Some NICs offer integratedfield-programmable gate arrays (FPGAs) for user-programmable processing of network traffic before it reaches the host computer, allowing for significantly reducedlatencies in time-sensitive workloads.[18] Moreover, some NICs offer complete low-latencyTCP/IP stacks running on integrated FPGAs in combination withuserspace libraries that intercept networking operations usually performed by theoperating system kernel; Solarflare's open-sourceOpenOnload network stack that runs onLinux is an example. This kind of functionality is usually referred to asuser-level networking.[19][20][21]
^Although other network technologies exist, Ethernet (IEEE 802.3) and Wi-Fi (IEEE 802.11) have achieved near-ubiquity as LAN technologies since the mid-1990s.
.jpg)
