What Is A QoS Quality of Service CISCO Benefits And Types

What is a QoS

The term Quality of Service (QoS) describes the methods and resources available to network managers for allocating traffic on a network. It entails specifying the activities a network device can do on a message from the time it enters until it leaves in order to control properties like loss, jitter, latency, and bandwidth. This makes it possible for networks to prioritize some types of traffic over others.

You can also read Hot Standby Router Protocol, How HSRP Works & Applications

How does Quality of Service Work

First-in, first-out (FIFO) queueing is typically used by switches and routers to process frames and packets; this method does not distinguish between different types of traffic. Per-hop behaviours (PHBs) are the various treatments that QoS systems apply to packets as they go through the network. These PHBs may involve changing packet header fields, delaying, or discarding them.

QoS tools control four essential aspects of network traffic:

Bandwidth: How much data can be sent by an interface in a second. QoS tools regulate which traffic has access to bandwidth and in what quantities.

Delay (Latency): The amount of time it takes to transmit info to the recipient. For time-sensitive packets, QoS can minimize latency.

Jitter: The change in packet latency. QoS technologies can reduce latency to enhance user experience, particularly for applications that operate in real-time.

Loss: Message loss rate, typically represented as a percentage. Packet loss is decreased using QoS technologies, especially for traffic that is time sensitive. Loss frequently happens when new packets are rejected and device queues fill up.

You can also read What Is First Hop Redundancy Protocols CISCO & FHRP History

Benefits of Quality of Service

In order to improve the end-user experience, QoS is used to control competition for network resources. Principal advantages consist of:

Improved Quality of Experience (QoE): Controlled bandwidth, delay, jitter, and loss provide users the impression that apps are of higher quality.

Prioritization of Critical Traffic: Provides the performance required for applications such as voice, video, and mission-critical data, even in the event of network congestion.

Better Resource Utilization: Through traffic flow management, QoS can assist WAN links and current network hardware operate as efficiently as possible, possibly avoiding expensive capacity upgrades.

Enhanced WAN Performance: Better quality for VoIP applications can be achieved by configuring WAN services that are QoS-capable, including MPLS and Ethernet WANs, to identify and prioritise particular traffic types.

Drawbacks of Quality of Service

Complexity: Given the abundance of tools and values, creating and executing a thorough QoS plan can be difficult, particularly in large organizations.

Resource Overhead: Certain QoS systems use more processing cycles and bandwidth than others, especially those that offer reliability (such as TCP’s error recovery).

Underutilization (in some cases): Backup queues are one example of a network device resource that may be underutilized during regular operation in particular QoS queuing techniques.

Internet Limitations: When using the Internet to access public cloud services, there are often no QoS assurances, which can result in increased packet loss, jitter, and delay, all of which can negatively affect the user experience.

You can also read What is Virtual Router Redundancy Protocol & How VRRP Works

Types of QoS Mechanisms

QoS tools specify the actions a device does in response to a message; each tool handles a particular issue and controls loss, jitter, latency, and bandwidth. Among the important mechanisms are:

Classification and Marking

The process of comparing packet header information to determine the types of traffic for QoS action is called classification. To match IP addresses, ports, and other criteria, ACLs (Access Control Lists) can be used.
Changing bit values in packet or frame headers to mark traffic for QoS treatment is known as marking.
- Differentiated Services Code Point (DSCP): The IPv4 Type of Service (ToS) byte or IPv6 Traffic Class field has a 6-bit field with 64 classifications. The DiffServ model’s essential component is this.
  - Expedited Forwarding (EF): Decimal 46 is the recommended DSCP number for low latency, jitter, and loss traffic, usually voice packets.
    - Assured Forwarding (AF): outlines a matrix of 12 DSCP values, each of which has three drop priority and four queuing classes.
    - Class Selector (CS): values for the DSCP that offer backward compatibility with the previous IP Precedence (IPP) values.
- Class of Service (CoS): The 802.1Q Ethernet header contains a 3-bit field called Priority Code Point (PCP), which is used for Layer 2 marking. It is only present when the frame is connected to trunk links.
- IP Precedence (IPP): the IPv4 ToS byte’s initial 3-bit field.
- MPLS EXP (Experimental) field: Currently used in MPLS headers for QoS marking, the Traffic Class field is the official name.
Trust Boundaries: The configuration of devices allows them to either “trust” or “untrust” other devices’ QoS markings. Usually, markings from untrusted domains are disregarded in order to avoid misuse.

Queuing (Congestion Management)

When an outgoing interface is busy, network devices temporarily store packets in queues.
FIFO (First-In, First-Out): No priority is given to packets; they are handled in the order they arrive.
Class-Based Weighted Fair Queuing (CBWFQ): Uses numerous queues and a round-robin scheduler to ensure that each class of traffic receives a certain percentage of the bandwidth. Usually used for data transmission, it offers fairness and bandwidth assurances but no latency guarantee.
Low Latency Queuing (LLQ): An improvement to CBWFQ that ensures minimal loss, jitter, and delay by establishing a priority queue for time-sensitive traffic (such as interactive video and audio). It assures bandwidth as well as latency.
Additional historical queuing techniques include Weighted Fair Queuing (WFQ), Custom Queuing (CQ), and Priority Queuing (PQ).

Policing

When traffic surpasses a predetermined policing rate, police officers either delete the extra communications or move them to a new class of service (often with a higher drop likelihood).
In order to enforce a committed information rate (CIR), policing is frequently employed at the edge of two networks, such as a WAN connection between a business and a service provider.

Shaping

Shaping keeps an eye on message rates as well, but instead of deleting unnecessary traffic, it keeps packets in queues and releases them gradually at a predetermined shaping rate.
Shapers can be set up to reduce this effect on voice and video traffic, however shaping’s side effect is greater jitter and latency because of queuing.

Congestion Avoidance

Through proactive packet discarding, particularly from TCP connections, these solutions seek to minimize overall packet loss before queues fill up entirely. By using this tactic, “tail drop,” which occurs when queues are fully filled, is less likely to occur.
Congestion avoidance techniques can take advantage of TCP’s windowing functionality, which causes TCP connections to slow down and reduces congestion by discarding some TCP segments when queue depths fall between minimum and maximum criteria.

Traffic Types and QoS Requirements

Different traffic types require different QoS:

Voice Applications (e.g., VoIP): Due to their susceptibility to loss, jitter, and delay, they demand stringent QoS. Cisco advises:

Delay (one-way): 150 ms or less.
Jitter: 30 ms or less.
Loss: 1% or less.
Voice traffic requires a consistent amount of bandwidth.

Video Applications: Like voice, but requiring more bandwidth and having a little higher threshold for jitter and delay:

Bandwidth: 384 Kbps to 20+ Mbps.
Delay (one-way): 200–400 ms.
Jitter: 30–50 ms.
Loss: 0.1%–1%.

Interactive Data Applications (e.g., web browsing): Can withstand higher levels of loss, jitter, and delay than voice/video, but QoE is still crucial for users. Data flow is often erratic.

Non-interactive Data Applications (e.g., data backup, file transfers): Jitter and delay don’t matter as much. Providing enough bandwidth and controlling loss are the main goals in order to lower the number of retransmissions.

Examples and Applications

Numerous network devices and situations use QoS technologies.

WAN Edge Routers: To control traffic leaving for slower WAN routes, queuing, policing, and shaping are frequently used.

Enterprise Networks:

For interactive audio and video, LLQ with a priority queue is a popular technique that achieves low delay, jitter, and loss. For data classes and non-interactive voice and video, CBWFQ (round-robin) is used, which provides more guaranteed bandwidth to business-critical data applications.
For applications like VoIP, where delay sensitivity is significant, QoS is essential to preventing service failure.

Service Provider Networks: Multi-Protocol Label Switching, or MPLS, was one of the first WAN services to provide useful QoS features. Call quality can be enhanced by customers marking VoIP packets (for example, using DSCP EF) so that the MPLS network treats them better.

Wireless Networks: To prioritise voice and video traffic, wireless LAN controllers (WLCs) provide QoS profiles with several queues (Bronze, Silver, Gold, and Platinum). An EasyQoS feature in Cisco DNA Centre makes it easier to set QoS policies by classifying apps into groups such as Default, Business Relevant, and Business Irrelevant.

You can also read Cisco Gateway load balancing protocol GLBP, How GLBP works

Page Content

Tutorials