Anetwork scheduler, also calledpacket scheduler,queueing discipline (qdisc) orqueueing algorithm, is anarbiter on anode in apacket switching communication network. It manages the sequence ofnetwork packets in the transmit and receivequeues of theprotocol stack andnetwork interface controller. There are several network schedulers available for the differentoperating systems, that implement many of the existing networkscheduling 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 oneflow,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 getsdropped.
A network scheduler may have responsibility in implementation of specificnetwork traffic control initiatives. Network traffic control is an umbrella term for all measures aimed at reducingnetwork congestion, latency and packet loss. Specifically,active queue management (AQM) is the selective dropping of queued network packets to achieve the larger goal of preventing excessive network congestion. The scheduler must choose which packets to drop.Traffic shaping smooths the bandwidth requirements of traffic flows by delaying transmission packets when they are queued in bursts. The scheduler decides the timing for the transmitted packets.Quality of service (QoS) is the prioritization of traffic based on service class (Differentiated services) or reserved connection (Integrated services).
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 receivebuffers.[1]Queuing disciplines are commonly used as attempts to compensate for various networking conditions, like reducing thelatency for certain classes of network packets, and are generally used as part of QoS measures.[2][3][4]
Classful queueing disciplines allow the creation of classes, which work like branches on a tree. Rules can then be set to filter packets into each class. Each class can itself have assigned other classful orclassless queueing discipline. Classless queueing disciplines do not allow adding more queueing disciplines to it.[5]
Examples of algorithms suitable for managing network traffic include:
| Algorithm | Acronym | Type | HW Support |
|---|---|---|---|
| Generic cell rate algorithm | GCRA | ||
| CHOose and Kill for unresponsive flows | CHOKe | Classless | |
| Controlled delay | CoDel | Classless | |
| Common Applications Kept Enhanced[6] | CAKE | ||
| Earliest TxTime First | ETF | Classless | Yes |
| First in, first out | FIFO | Classless | |
| Fair queuing | FQ | Classless | |
| Fair Queuing Controlled Delay | FQ-CoDel | Classless | |
| Flow Queuing with Proportional Integral controller Enhanced | FQ-PIE | Classless | |
| Generalized Random Early Detection | GRED | Classless | |
| Heavy-Hitter Filter[7] | HHF | Classless | |
| Multiqueue Priority | MQ-PRIO | Classless | Yes |
| Multiqueue | MULTIQ | Classless | Yes |
| Network Emulator[8] | NETEM | Classless | |
| Proportional Integral controller-Enhanced[9] | PIE | Classless | |
| Random early detection | RED | Classless | |
| Stochastic fair Blue | SFB | Classless | |
| Stochastic Fairness Queueing | SFQ | Classless | |
| Token Bucket Filter | TBF | Classless | |
| Class-based queueing | CBQ | Classful | |
| Credit-Based Shaper | CBS | Classful | Yes |
| Deficit round robin[10] | DRR | Classful | |
| Enhanced Transmission Selection | ETS | Classful | |
| Hierarchical fair-service curve | HFSC | Classful | |
| Hierarchical Token Bucket[11] | HTB | Classful | |
| Priority | PRIO | Classful | |
| Quick Fair Queueing[12] | QFQ | Classful | |
| Time Aware Priority Shaper | TAPRIO | Classful | Yes |
Several of the above have been implemented asLinux kernel modules[13][14] and arefreely available.
Bufferbloat is a phenomenon in packet-switched networks in which excessbuffering of packets causes highlatency andpacket delay variation. Bufferbloat can be addressed by a network scheduler that strategically discards packets to avoid an unnecessarily high buffering backlog. Examples includeCoDel,FQ-CoDel andrandom early detection.
![[icon]](/mt/?noimg=&dark=on&url=https%3a%2f%2fen.wikipedia.org%2fwiki%2fFile%3aWiki_letter_w_cropped.svg) | This sectionneeds expansion. You can help byadding missing information.(October 2018) |
The Linux kernel packet scheduler is an integral part of theLinux kernel's network stack and manages the transmit and receivering buffers of all NICs.
The packet scheduler is configured using the utility calledtc (short fortraffic control). As the default queuing discipline, the packet scheduler uses a FIFO implementation calledpfifo_fast,[15] althoughsystemd since its version 217 changes the default queuing discipline tofq_codel.[16]
Theifconfig andip utilities enable system administrators to configure the buffer sizestxqueuelen andrxqueuelen for each device separately in terms of 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.[a]
Berkeley Packet Filter filters can be attached to the packet scheduler's classifiers. TheeBPF functionality brought by version 4.1 of the Linux kernel in 2015 extends the classic BPF programmable classifiers to eBPF.[17] These can be compiled using theLLVM eBPF backend and loaded into a running kernel using thetc utility.[18]
ALTQ is the implementation of a network scheduler forBSDs. As of OpenBSD version 5.5 ALTQ was replaced by the HFSC scheduler.
Schedulers in communication networks manage resource allocation, including packet prioritization, timing, and resource distribution. Advanced implementations increasingly leverage artificial intelligence to address the complexities of modern network configurations. For instance, a supervised neural network (NN)-based scheduler has been introduced in cell-free networks to efficiently handle interactions between multiple radio units (RUs) and user equipment (UEs). This approach reduces computational complexity while optimizing latency, throughput, and resource allocation, making it a promising solution for beyond-5G networks.[19]
