Network scheduler

Packets queuing in a FIFO data structure.

A network scheduler, also called packet scheduler, is an arbiter program that manages the sequence of network packets in the transmit and receive queues of the network interface controller, which is a circular data buffer. There are several network schedulers available for the different operating system kernels, that implement many of the existing network scheduling algorithms.

Also, network schedulers are enabling accomplishment of the active queue management and traffic shaping.

Implementations

Linux kernel

The LAMP (software bundle) (here additionally with Squid) is a high performance and high-availability heavy duty solution for a hostile environment. As part of the Linux kernel, the Linux Network Scheduler has to scale and deliver performance

The currently available network scheduler for the Linux kernel is an integral part of the Linux network stack; it arbiters the ringbuffers of all NICs, hence it works on layer 2 of the OSI model and handles Ethernet frames. It manages the transmit and receive buffers of all present NICs.

The user space programs ifconfig and ip enable the system administrator to regulate the buffer sizes: txqueuelen /rxqueuelen; the unit being number of Ethernet frames (regardless of their size) for each device separately. The Linux 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.

Another network scheduler is being developed as part of Netfilter.

BSD

ALTQ is the implementation of a network scheduler for Berkeley Software Distribution.

Functioning

The network scheduler logic decides similar to a statistical multiplexer which network packet to forward next from the buffer. The buffer works as a queuing system, where the network packets are stored temporarily until they are transmitted. The buffer space is divided into many queues, each of which is used to hold the packets of one flow, defined for instance by source and destination IP addresses.

The in each case loaded network scheduling algorithm and the settings determine how the network scheduler manages the buffer.

Bufferbloat

High latency and jitter are phenomenons encountered due to the coincidence of bufferbloat and high traffic volume. Though the network scheduler manages only the transmit (and receive) queue of the network controller, and more buffers exist, by the adoption of the CoDel queueing disclipine, a significant reduction of the effect has been achieved.

Network scheduling algorithms

In the course of time several network scheduling algorithms (also called queueing discipline or queueing algorithm) have been developed. There is a rough distinction between leaky bucket and token bucket algorithms, and between class-based and class-less ones. The head drop discipline CoDel is less prone to the effects of bufferbloat then the common tail drop disciplines.

Several scheduling algorithms suited for the task of managing network packets have been implemented as Linux kernel modules^[1] and are freely available:

Buzzwords

Though the task of a network scheduler is not that complicated, a lot of buzzwords are around:

• 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 – usually used when more individuals share the same Internet connection
• Traffic Prioritizing – a form of AQM, that selective prioritizes certain network packets (e.g. VoIP-packets)
• TCP Turbo – winner of the buzzword contest, accentuates, that by prioritizing ACK packages on the upload, the download rate of a TCP connection is not slowed down
• Bandwidth Management – the management of the available bandwidth

See also

References