QoS Configuration: Traffic Priority for VoIP and Video Conferencing

TL;DR: QoS (Quality of Service) configuration — traffic prioritisation in the SME office for VoIP and video conferencing, with a hands-on guide.
Summary: QoS (Quality of Service) is the discipline of assigning different priorities to different traffic types on a network. In an SME office, limited internet bandwidth is split between VoIP calls, Teams / Zoom meetings, file transfers, web browsing, and YouTube; without QoS, the noisiest application (e.g. YouTube) hurts the most important one (VoIP). Properly configured QoS lifts VoIP to the highest priority, puts video conferencing second, file transfers third, and pushes "background" traffic to the back. It's applied on switches and routers via DSCP marking; at SME scale it can be set up in an afternoon and runs quietly for years.
In SMEs, the "internet is slow" complaint usually isn't about raw bandwidth — it's about traffic mix. On a single link, 50 devices do different things at once: accounting pushes data into the ERP, sales is on a Teams call, the office is on Zoom, someone is on Netflix during a break, the cloud backup started its nightly batch two hours early. Without QoS, all that traffic carries the same priority — and the customer's VoIP call breaks up.
In this article we cover SME-scale QoS configuration, the VoIP / video prioritisation strategy, and switch / router-level implementation. Target audience: IT managers, network administrators, and managers who keep getting complaints about VoIP / video-conference quality.
Why QoS Is Necessary
The internet / LAN runs "best effort" — packet order and delay aren't guaranteed. QoS overlays a prioritisation rule on that default.
An Environment Without QoS
[VoIP packet] | [Backup packet] | [VoIP packet] | [Backup packet]
queue
Backup is large (1500 bytes), VoIP is small (~200 bytes) and needs regular intervals. If backup packets fill the queue, VoIP gets delayed and breaks up.
An Environment With QoS
[VoIP packet] | [VoIP packet] | [Backup packet] | [Backup packet]
high priority low priority
VoIP always goes first. Backup isn't hurt by waiting (no human is watching).
Classifying Traffic
QoS needs traffic split into categories.
Typical QoS Classes
| Class | Application | Latency | Bandwidth |
|---|---|---|---|
| Voice (Real-Time) | VoIP, IP phones | <150 ms | Low (~100 kbps) |
| Video (Real-Time) | Teams, Zoom, Meet | <150 ms | High (1–5 Mbps) |
| Critical Data | ERP, CRM, corporate portal | <500 ms | Medium |
| Important Data | Email, file transfer | <1 second | Medium-high |
| Best Effort | Web browsing | Tolerant | |
| Bulk | Backup, OS update | Doesn't matter | Low priority |
| Scavenger | YouTube, social media | Doesn't matter | Lowest |
A Lean Model for SMEs
Three classes can be enough:
- High: VoIP, video conferencing
- Medium: Business-critical applications (ERP, M365)
- Low: Web, backup, entertainment
DSCP Marking
QoS rollout uses the DSCP (Differentiated Services Code Point) standard.
Typical DSCP Values
| DSCP | Name | Recommended use |
|---|---|---|
| EF (46) | Expedited Forwarding | VoIP |
| AF41 (34) | Assured Forwarding | Video conferencing |
| AF31 (26) | Critical Apps | ERP, business-critical |
| AF21 (18) | Important | |
| 0 | Default (Best Effort) | Web browsing |
| CS1 (8) | Scavenger | YouTube, personal |
DSCP is a field in the IP header — wherever the packet goes, it carries the signal "I'm high priority".
Where to Mark
DSCP marking can happen in three places:
- In the application: the VoIP software marks its own packet as EF
- On the switch: trust port (accepts the application's marking)
- On the router: mark outbound traffic before it hits the WAN
Typical SME approach: VoIP / video applications set their own markings, the switch trusts them, the router / firewall sends them out prioritised onto the WAN.
QoS Architecture at SME Scale
QoS Inside the LAN
On the switch side:
- Voice VLAN (for VoIP devices)
- DSCP trust enabled
- Queue priority (4–8 queues depending on hardware)
- Strict priority (the highest queue goes first)
QoS Toward the WAN
On the router / firewall:
- Outbound shaping (shape traffic to WAN bandwidth)
- Class-based queuing
- Bandwidth guarantees (e.g. 10% for VoIP)
- Maximum caps (Scavenger limited to 20%)
Controlling Inbound
Inbound QoS is limited — without ISP-side QoS you don't have full control. But:
- BBR / CUBIC congestion control
- Application-aware shaping
- More advanced in SD-WAN setups
QoS for VoIP — A Practical Example
VoIP is the most sensitive traffic type.
VoIP Requirements
- Latency: <150 ms one-way
- Jitter: <30 ms
- Packet loss: <1%
- Bandwidth (G.711 codec): 87 kbps one-way
VoIP QoS Configuration
- A separate VLAN for VoIP devices (phones, softphones)
- Default DSCP EF (46) on the Voice VLAN
- Switch port: trust dscp / trust voice
- Voice queue on the router: strict priority
- Bandwidth: at least 10–15% guaranteed for VoIP
Example Cisco Switch Commands
interface GigabitEthernet1/0/1
switchport mode trunk
switchport voice vlan 40
switchport access vlan 20
mls qos trust device cisco-phone
mls qos trust dscp
QoS for Video Conferencing
Applications like Teams, Zoom, Google Meet.
Requirements
- Latency: <150 ms
- Jitter: <30 ms
- Packet loss: <1% (audio), <1.5% (video)
- Bandwidth: 1–5 Mbps one-way (HD)
Configuration
- Microsoft 365: Teams traffic uses DSCP 46/34/24 (Microsoft's guidance)
- Zoom: DSCP 46 (audio), 34 (video)
- For configuration, consult the Microsoft Teams QoS documentation
A Common Issue
Cloud SaaS applications (Teams, Zoom) may have the DSCP marker rewritten on the provider side, and ISPs may zero out DSCP. In that case, DSCP marking on the SME side has limited effect; but it still improves things inside the local network.
Bandwidth Management
QoS is not just about priority; it's also about bandwidth planning.
A Typical SME Pipe
- 100/100 Mbps fibre
- 30 employees
- Active use: 50 devices
Typical Consumption
- Background (OS updates, downloads): 30–40%
- Web browsing, email: 20–30%
- Cloud apps (M365, Salesforce): 15–25%
- Teams / Zoom video: 20–40% (with many participants)
- VoIP: 1–5%
Total: can exceed 100% → congestion → user complaints.
Resolution
Managed prioritisation via QoS + (if needed) more bandwidth. Just adding bandwidth without QoS doesn't fix it — someone on Netflix can consume 50 Mbps; you need the right limits.
Traffic Shaping
A slightly different concept from QoS: capping bandwidth.
Typical Uses
- Guest Wi-Fi: 10 Mbps max per user
- A specific IP / device: 20 Mbps max
- A specific application: 30 Mbps max (e.g. YouTube)
Configuration
In modern firewalls (FortiGate, Palo Alto, Sophos):
- Application-aware shaping
- Limits by IP / user / group
- Schedules (strict during working hours, looser outside)
A Typical SME Policy
| User / Traffic | Limit |
|---|---|
| Guest Wi-Fi | 10 Mbps/user |
| Employee total | 50 Mbps/user |
| Streaming (YouTube, Netflix) | 20% of total bandwidth |
| Backup traffic | Wide overnight, narrow during the day |
| OS updates | Overnight only |
Testing and Verification
How do you check QoS is doing its job?
Tools
- Wireshark: to see whether DSCP markers are correct
- iperf3: bandwidth tests
- Pingplotter / Smokeping: latency / jitter monitoring
- MOS score: VoIP-quality measurement (automatic on VoIP gateways)
KPIs
- VoIP MOS score: should be >4.0
- Latency (inside LAN): <5 ms
- Latency (WAN): <50 ms (local), <150 ms (long-distance)
- Jitter: <30 ms
- Packet loss: <1%
Common QoS Mistakes
Typical issues in SMEs:
- I'm marking DSCP but trust is missing: if the switch port doesn't trust, the marking is overwritten
- Everything's marked high priority: nothing is
- No inbound QoS: download congestion can't be eased
- No VLAN separation: voice traffic mixes with data
- No bandwidth shaping: background traffic balloons
- Not reviewed in years: new applications aren't in the QoS plan
- SaaS DSCP isn't preserved: rewritten in the cloud — but marking still helps inside the LAN
What Yamanlar Bilişim Offers
Our QoS support areas at SME scale:
- Current traffic analysis (NetFlow, sniffing)
- VoIP / video quality measurement
- QoS policy design
- Switch and router configuration
- A DSCP-marking standard
- Bandwidth planning
- Annual QoS review
- VoIP MOS monitoring
Frequently Asked Questions
Conclusion
QoS is the quiet discipline that prevents VoIP, video conferencing, and business-critical traffic from slowing down in a bandwidth-constrained environment. At SME scale, DSCP marking + switch/router configuration + an annual review goes up in an afternoon and runs in the background for years. "Slow internet" complaints are often caused by missing QoS, not insufficient bandwidth.
Yamanlar Bilişim provides traffic analysis, QoS design, and rollout services sized to your needs — moving the quality of customer calls in your office from "good if we're lucky" to confidently guaranteed.
Frequently Asked Questions
Is QoS really necessary in an SME office?
If VoIP or video conferencing is heavily used — absolutely yes. In an office that only browses the web and does email, QoS yields less, but doesn't hurt. With 5+ people on continuous Teams/Zoom + VoIP, quality complaints without QoS are inevitable.
Does QoS help for cloud-SaaS (Teams, Zoom) traffic?
Limited. On the SME → ISP → SaaS path, the DSCP marker is frequently zeroed out. But it still helps within the local network (up to the ISP); when the pipe is congested, cloud traffic gets prioritised. Microsoft recommends QoS configuration for Teams.
Which switch meets QoS requirements for an SME?
A Layer 2+ managed switch (e.g. Cisco Catalyst, Aruba Instant On, Ubiquiti UniFi, MikroTik) supports most QoS features. Layer 3 switches are preferred (VLAN routing). An unmanaged switch or a home router can't do QoS.
Is putting VoIP on a separate link better than QoS?
Possible, but it raises cost. For an SME, a single link with good QoS is more economical and sufficient than a dual link. Dual links make sense for environments with critical VoIP needs (call centres, continuous customer phone work). For a standard SME office, good QoS is enough.
Does QoS actually reduce bandwidth?
No — it just shares it . When there's congestion, prioritised traffic gets through; the rest waits. With no congestion, QoS is invisible — everyone uses full speed. So in normal SME use, QoS is unnoticed; during rush hour, it saves the day.
Is NetFlow / sFlow traffic analysis mandatory?
Helpful for designing QoS policy, not mandatory. Manually: which apps are heavy? Ask employees, read logs. Automated: NetFlow / sFlow + ntopng / SolarWinds NTA — visualise the real distribution. Small SMEs can do without; mid-scale and up, NetFlow is a practical investment.
Author
Serdar
Yamanlar Bilişim Expert
Writes content on IT infrastructure, cybersecurity, and digital transformation at Yamanlar Bilişim. Get in touch for any questions.
Professional Support
Get help on this topic
Let's design the Network and Security solution you need together. Our experts get back to you within 1 business day.
support@yamanlarbilisim.com.tr · Response time: 1 business day
Keep Reading
Related Articles

Getting Ready for IPv6: When and How Should an SME Make the Move?
What IPv6 is, when an SME should make the move, dual-stack architecture, and a practical preparation guide.

Managing Guest Wi-Fi with a Captive Portal
What a captive portal is, how it's deployed in SME offices and guest-Wi-Fi scenarios, Law-5651-compliant logging, and brand-experience guide.

Moving to Wi-Fi 6 and 6E: Coverage Planning for an SME Office
Wi-Fi 6 (802.11ax) and Wi-Fi 6E features — the SME-office migration decision, coverage planning, and device-compatibility guide.