Network scheduler

Packets queuing in a FIFO (first in, first out) data structure.

A network scheduler, also called packet scheduler, queueing discipline, qdisc or queueing algorithm, is an arbiter on a node in packet switching communication network. It manages the sequence of network packets in the transmit and receive queues of the network interface controller. There are several network schedulers available for the different operating systems, that implement many of the existing network scheduling algorithms.

The network scheduler logic decides which network packet to forward next. The network scheduler is associated with a queuing system, storing the network packets temporarily until they are transmitted. Systems may have a single or multiple queues in which case each may hold the packets of one flow, classification, or priority.

In some cases it may not be possible to schedule all transmissions within the constraints of the system. In these cases the network scheduler is responsible for deciding which traffic to forward and what gets dropped.

Algorithms

In the course of time many network queueing disciplines have been developed. Each of these provides specific reordering or dropping of network packets inside various transmit or receive buffers.^[1][2] Queuing disciplines are commonly used as attempts to compensate for various networking conditions, like reducing the latency for certain classes of network packets, and are generally used as part of quality of service (QoS) measures.^[3][4][5]

Examples of algorithms suitable for managing network traffic include:

Several of the above have been implemented as Linux kernel modules^[17] and are freely available.

Bufferbloat

Bufferbloat is a phenomenon in packet-switched networks generally, in which excess buffering of packets causes high latency and packet delay variation (also known as jitter), as well as reducing the overall network throughput. When a router device is configured to use excessively large buffers, even very high-speed networks can become practically unusable for many interactive applications like voice calls, chat, and even web surfing.

The CoDel algorithm attempts to reduce this problem by improving upon the RED algorithm. CoDel is less prone to the effects of bufferbloat than the common tail drop disciplines.

Terminology

• Active queue management (AQM) – the employment of an arbiter program
• Network traffic control – an umbrella term for all measures aimed at the control of traffic
• Traffic shaping – a form of AQM, where the available bandwidth is being limited to match certain use cases
• Link sharing – this term is used when more individuals or customers share the same Internet connection
• Traffic prioritizing – a form of AQM, that selectively prioritizes certain network packets (e.g. VoIP-packets)
• TCP Turbo – by prioritizing ACK-packets on the upload, a slow-down of the download rate of a TCP connection is prevented
• Bandwidth management – the management of the available bandwidth

Implementations

Linux kernel

The Linux kernel's packet scheduler is part of the network stack, together with netfilter, nftables, and Berkeley Packet Filter.

The Linux kernel packet scheduler is configured using the userspace CLI utility called tc (short for "traffic control"). As the default queuing discipline, the packet scheduler uses a FIFO implementation called pfifo_fast,^[18] although systemd since its version 217 changes the default queuing discipline to fq_codel.^[19] The packet scheduler is an integral part of the Linux kernel's network stack and manages the ring buffers of all NICs, by working on the layer 2 of the OSI model and handling Ethernet frames, for example. It manages the transmit and receive buffers of all NICs installed in a computer.

The user space programs ifconfig and ip enable system administrators to configure the buffer sizes txqueuelen and rxqueuelen for each device separately, with their units being number of Ethernet frames regardless of their size. The Linux kernel's network stack contains several other buffers, which are not managed by the network scheduler. The overall size of all buffers has been the point of critique by the Bufferbloat project, which provided a partial solution with CoDel that has been primarily tested in OpenWrt.

Another network scheduler is being developed as part of Netfilter and nftables.

eBPF was merged into the Linux kernel mainline in kernel version 3.18, and eBPF filters can also be attached to the packet scheduler's classifiers since the kernel version 4.1.^[20] The eBPF functionality brought by version 4.1 of the Linux kernel extends the "classic" BPF programmable classifier by extending its scope to native eBPF code, allowing userspace to implement its own custom, safe, C-like classifiers that can be compiled using the LLVM eBPF backend and loaded into a running kernel using the tc utility.^[21]

OpenBSD

Since OpenBSD version 5.5 ALTQ was totally replaced by HFSC scheduler.

BSD

ALTQ is the implementation of a network scheduler for BSDs.

See also

• Network congestion
• Quality of service
  • Differentiated services
  • Integrated services
• Queue (abstract data type)
• Queueing theory
• Statistical time division multiplexing
• Traffic shaping
• Traffic classification
• Type of service

References