Understanding Internet QoS and Traffic Scheduling for Better Network Performance

Why Is Traffic Scheduling Important?

Imagine you are at an airport security checkpoint. Passengers with first-class tickets (VoIP packets) are allowed through first, followed by business class (video), and finally economy class (best-effort traffic like email). This orderly handling is similar to what happens in networks using traffic scheduling.

In a network context, packets are categorized based on their QoS class. For example:

• Red packets: High-priority voice traffic
• Green packets: Medium-priority video streams
• Blue packets: Low-priority FTP or HTTP traffic

Traffic scheduling ensures each category is processed appropriately, with higher priority packets being sent faster and more reliably than others.

Packet Classification and Service Level Agreements (SLAs)

The foundation of traffic scheduling is packet classification and marking. This involves:

• Identifying the type of traffic (e.g., VoIP, YouTube, FTP)
• Assigning it a QoS class based on Service Level Agreements (SLAs)

For example, if your mobile plan includes VoIP services, your provider may prioritize your voice packets at the base station itself. This application-level SLA ensures voice traffic gets precedence over social media or downloads.

Types of Traffic Classes

1. High-Priority Delay-Sensitive Traffic

   • Example: VoIP, online gaming
   • Needs: Minimal queuing delay and jitter

2. Medium-Priority Bandwidth-Hungry Traffic

   • Example: Video-on-demand, IPTV
   • Needs: High bandwidth, low jitter

3. Low-Priority Best-Effort Traffic

   • Example: Email, web browsing
   • Needs: No strict QoS guarantees

Multi-Class Scheduling: Ensuring Fairness and Priority

To handle diverse traffic needs, networks use multi-class scheduling. This involves maintaining separate queues for each traffic class and employing distinct scheduling strategies for each.

Key Goals:

• Reduce queuing delays for high-priority traffic
• Ensure adequate bandwidth for video streams
• Provide best-effort service for low-priority traffic

Let’s explore popular queuing strategies used in traffic scheduling.

1. Priority Scheduling

Non-Preemptive Priority Scheduling

• Packets in the high-priority queue are served first.
• Once that queue is empty, the scheduler serves the next priority queue, and so on.
• The process follows a round-robin among queues but without interrupting once a queue’s turn begins.

Pros:

• Simplicity
• Predictable behavior for high-priority traffic

Cons:

• Lower-priority queues might face starvation during high traffic loads

Preemptive Priority Scheduling

• Higher-priority packets can interrupt ongoing transmissions
• If a new VoIP packet arrives while video traffic is being served, the scheduler immediately switches back to serving VoIP

Pros:

• Excellent for time-sensitive applications
• Ensures strict QoS for VIP packets

Cons:

• Risk of starvation for lower-priority traffic

2. Custom Queuing (CQ)

Custom queuing allows the definition of queue size percentages, e.g.:

• Queue 1 (VoIP): 30% of buffer capacity
• Queue 2 (Video): 20%
• Queue 3 (FTP): 50%

How It Works:

• Each queue is served in round-robin order.
• During congestion, packets from queues with smaller sizes are more likely to be dropped.
• This method guarantees bandwidth allocation and is effective when traffic volumes are high.

Use Case: Excellent for applications that demand guaranteed bandwidth, like corporate video conferencing or dedicated streaming services.

3. Weighted Fair Queuing (WFQ)

WFQ ensures fair bandwidth distribution across traffic classes, even when packet sizes vary.

Key Idea:

• Calculate weights based on packet size and desired fairness
• For example, if:
  • Blue packets are 1 unit
  • Red packets are 4 units
  • Green packets are 2 units
  Then, serving 4 blue, 1 red, and 2 green packets ensures equal bandwidth distribution.

Benefit:

• Prevents starvation
• Maintains fairness while allowing priority enforcement

Multilevel Queue Scheduling

In real-world systems, multiple queuing techniques are often combined:

• Level 1: Priority scheduling (high-level sorting)
• Level 2: WFQ within each priority class (to handle variable packet sizes fairly)

This layered approach balances the need for both priority enforcement and fairness, offering an optimized QoS strategy.

Congestion Avoidance vs. Congestion Control

While TCP handles congestion control by reducing its transmission rate upon detecting packet loss, congestion avoidance attempts to prevent congestion proactively—especially for inelastic traffic like video streams.

Why It Matters:

• Real-time applications can’t afford delays introduced by TCP’s control mechanisms.
• Without congestion avoidance, elastic TCP flows might overwhelm inelastic traffic like UDP video, causing severe packet drops.

RED (Random Early Detection): The Key to Congestion Avoidance

RED is a proactive method used to detect and prevent congestion before it happens.

How RED Works:

1. Measure average queue length
2. Define thresholds:
   • If average queue < minimum threshold → enqueue packet
   • If in between min and max thresholds → calculate packet drop probability
   • If above max threshold → drop packet

Why "Random"? To avoid synchronized packet drops that could lead to cascading TCP slowdowns, RED uses random drops based on calculated probabilities.

This equation ensures gradual response to growing congestion and avoids sudden traffic halts.

Why Congestion Avoidance and Control Must Coexist

RED protects high-priority flows by keeping queues balanced. But when congestion becomes unavoidable:

• TCP’s congestion control kicks in for elastic flows like FTP
• RED ensures inelastic flows like VoIP remain unaffected

Together, they maintain a balanced and QoS-compliant network.

Final Thoughts

Understanding traffic scheduling and congestion avoidance is vital for implementing QoS in modern networks. From VoIP to video streaming, from corporate traffic to cloud gaming, QoS mechanisms ensure applications receive the performance they require.

If you're tackling assignments on these topics or preparing for exams, don’t hesitate to use our computer network assignment help. Our experts are here to guide you through everything from RED algorithms to WFQ models with practical insights and academic rigor.