What Is QoS (Quality of Service)?
Table of Contents
What QoS Does
Quality of Service (QoS) is a set of techniques for prioritizing certain types of network traffic over others. When a network link is congested, QoS determines which packets get forwarded first and which ones wait (or get dropped).
Without QoS, all traffic is treated equally. A video call competes with a file download competes with a backup job. If the link is saturated, all three degrade equally. With QoS, the video call's packets get priority because real-time communication is more sensitive to delay than a file transfer that can tolerate buffering.
Why QoS Exists
Different applications have different tolerance for delay, jitter (variation in delay), and packet loss:
- Voice/video calls - need low latency (under 150ms), low jitter, can tolerate small packet loss (codecs handle it)
- Web browsing - tolerates moderate latency, needs reliable delivery
- File transfers - tolerate high latency, need complete delivery but timing is flexible
- Backups/updates - lowest priority, should use only unused capacity
QoS lets the network match its behavior to these different requirements rather than treating a VoIP packet the same as a Windows Update download.
graph TD
subgraph "QoS Priority Queues"
VOICE["Voice (AC_VO)
Highest priority"] --> OUT["Network Output"]
VIDEO["Video (AC_VI)
High priority"] --> OUT
BEST["Best Effort (AC_BE)
Normal priority"] --> OUT
BG["Background (AC_BK)
Lowest priority"] --> OUT
end
QoS queuing: packets are classified into priority tiers. Voice gets transmitted first, background traffic waits.
How QoS Works
QoS operates through three mechanisms:
Classification - identifying what type of traffic a packet belongs to. This can be done by port number (port 5060 = SIP/VoIP), protocol, source/destination address, or DSCP markings already in the packet header.
Queuing - placing classified packets into different priority queues. High-priority queues are serviced first. Common queuing algorithms include strict priority (high queue always goes first), weighted fair queuing (each queue gets a proportional share), and low-latency queuing (combination of both).
Shaping/policing - limiting the rate of certain traffic classes to prevent them from consuming all bandwidth. Shaping buffers excess traffic and releases it gradually. Policing drops excess traffic immediately.
DSCP Marking
The Differentiated Services Code Point (DSCP) is a 6-bit field in the IP header that marks packets with their priority class. Routers read the DSCP value to determine how to handle the packet.
Common DSCP values:
- EF (Expedited Forwarding, 46) - highest priority, low latency. Used for VoIP.
- AF (Assured Forwarding, 10-38) - multiple priority tiers with drop precedence. Used for business-critical applications.
- CS (Class Selector, 0-56) - backward-compatible priority levels. CS0 is best-effort (default).
- 0 (Best Effort) - no priority, the default for all unmarked traffic.
DSCP markings only matter if every router along the path respects them. On the public internet, ISPs typically ignore or re-mark DSCP values. QoS is most effective within a managed network where the administrator controls all routers.
WMM: QoS for WiFi
WiFi Multimedia (WMM) brings QoS to wireless networks. WMM defines four access categories with different priority levels for channel access:
- Voice (AC_VO) - highest priority, shortest contention window
- Video (AC_VI) - high priority
- Best Effort (AC_BE) - normal priority, default for most traffic
- Background (AC_BK) - lowest priority, longest contention window
Higher-priority categories get shorter CSMA/CA backoff times, meaning they win channel access more often. A voice packet waits less time before transmitting than a background download packet. WMM is mandatory in WiFi 6 (802.11ax) and supported in most modern devices.
QoS in Practice
QoS is most visible in enterprise and home networks where bandwidth is limited. A home router with "gaming mode" QoS prioritizes game traffic over other downloads. An office network with QoS ensures video conferences do not stutter when someone starts a large file transfer.
When the BLEShark Nano scans a WiFi network, it can observe the WMM parameters advertised in beacon frames. These parameters reveal how the AP prioritizes different traffic classes. A network with aggressive voice priority settings suggests VoIP deployment. Default WMM settings suggest the administrator has not customized QoS, which may mean all traffic competes equally during congestion.