Skip to main content

Quality of Service (QoS) - Key concepts

 Quality of Service Overview

QoS is crucial for managing bandwidth, latency, jitter, and packet loss, which impact application performance, especially for real-time services like voice and video. Key quality issues include lack of bandwidth, latency (packet delay), jitter (variability in delay), and packet loss (lost data during transmission). Technical Tip: Keep latency under 150 ms for real-time traffic and packet loss below 1%.

End-to-End Layer 3 QoS using MQC

DiffServ QoS Model

DiffServ is scalable and classifies traffic into classes, marking it for varying levels of priority. Technical Tip: Use DiffServ for efficient traffic management at the network edge for classification and marking, while relying on Per-Hop Behavior (PHB) in the core network.

CoS and DSCP Mapping

CoS marks traffic in Layer 2 headers, while DSCP marks it at Layer 3 for end-to-end traffic prioritization. Technical Tip: DSCP is more versatile for end-to-end QoS as CoS is limited to Layer 2 domains.

Classification and Marking

Classification identifies traffic based on multiple factors (Layer 2 to Layer 7), and marking assigns a QoS value for treatment. Technical Tip: Always classify and mark traffic close to the network ingress to optimize bandwidth.

Network-Based Application Recognition (NBAR)

NBAR classifies applications using Layer 7 inspection. Technical Tip: Use NBAR with Modular QoS CLI (MQC) for advanced traffic classification, and update NBAR with Packet Description Language Modules (PDLM) for new protocols.

Policing and Shaping

Policing drops or marks traffic that exceeds bandwidth limits, while shaping buffers excess traffic to smooth it out. Technical Tip: Use shaping for outbound traffic at network edges to comply with service provider limits, and policing to enforce strict traffic limits. Example:

policy-map POLICING_POLICY

 class CLASS1

  police 1000000 20000 exceed-action drop

Congestion Management and Avoidance

Congestion management uses queuing to prioritize traffic (e.g., CBWFQ), while avoidance tools like WRED preemptively drop packets to prevent congestion. Technical Tip: Use WRED to reduce tail drops and prevent global synchronization issues.

Queueing Algorithms

Different queuing methods include Priority Queueing (PQ), which gives absolute priority, and Class-Based Weighted Fair Queuing (CBWFQ), which allocates bandwidth to multiple queues. Technical Tip: Be cautious with PQ as it can starve lower-priority traffic.

Auto QoS

AutoQoS simplifies the setup of QoS for voice and video applications by automating configurations. Technical Tip: Enable AutoQoS before manual QoS setup to ensure proper interface settings. Use:

interface gigabitethernet0/1

 auto qos voip trust

Advanced QoS Concepts

RSVP

RSVP reserves bandwidth for real-time applications like voice and video. Technical Tip: Implement RSVP for applications that need guaranteed bandwidth, marking traffic with DSCP for Expedited Forwarding (EF).

Hierarchical QoS (HQoS)

HQoS manages traffic across different levels of the network, such as service provider and enterprise levels. Technical Tip: Use HQoS to manage traffic for different service levels efficiently, especially in environments with multiple service agreements.

QoS Configuration Example

class-map VOICE

 match ip rtp 16384 16383

policy-map VOICE_POLICY

 class VOICE

  set dscp ef

  priority 300

 class class-default

  bandwidth 1000

Key QoS Technical Tips

Classification and marking should be done at the network ingress. Use shaping to buffer excess traffic and avoid packet loss, and employ policing for strict rate enforcement. WRED prevents tail drops and mitigates global synchronization problems in congestion.

Comments

Popular posts from this blog

Basic MPLS BGP and L3VPN Lab Setup

In this lab, we’ve set up a basic MPLS, BGP, and L3VPN environment, which is a great foundation for understanding how service providers build scalable networks. The lab uses the EVE-NG simulator along with Router IOS C7200-ADVENTERPRISEK9-M, Version 15.2(4)M8 to emulate a realistic MPLS environment. Below is a summary of the key components and roles of each router in the lab. MPLS Core Routers : The MPLS core consists of the routers responsible for label switching and forwarding customer traffic through the network: PE1 (Provider Edge 1) : Connects customer networks to the MPLS core and handles both MPLS and BGP routing. It also hosts VRF (Virtual Routing and Forwarding) instances for customers. PE2 (Provider Edge 2) : Functions similarly to PE1, connecting another customer network to the MPLS core. P1 (Core Router 1) and P2 (Core Router 2) : These routers serve as MPLS core routers and handle label switching but do not store or process customer routes directly. They simply f

OSPF Adjacency Stuck in EXSTART on Cisco IOS XR – Issue and Solution

In a recent lab setup using Cisco IOS XR on EVE-NG, I faced a common but frustrating issue with OSPF adjacencies getting stuck in the EXSTART state. After spending considerable time troubleshooting interface MTUs and configurations, I discovered that the root cause was related to the virtual network interface type being used. This post outlines the issue, troubleshooting steps, and the eventual solution that got everything working. Issue: While configuring OSPF between two routers running Cisco IOS XR in my lab, OSPF adjacencies were getting stuck in the EXSTART state. I verified that interface configurations, MTU settings, and OSPF parameters were correct, but the problem persisted. I tried adjusting the MTU size, using the mtu-ignore command, and even checked for ACLs, but nothing seemed to resolve the issue. Troubleshooting Steps: MTU Settings: I started by verifying that both sides of the OSPF adjacency had matching MTUs. I used the default MTU and even tried different values wit

How to Properly Clone an EVE-NG Lab with Configurations

Cloning labs in EVE-NG is a great way to duplicate setups and expand or experiment on a new copy without affecting the original lab. However, if not done correctly, the cloned lab may only copy the topology without configurations. In this guide, I’ll show you how to properly clone a lab in EVE-NG with all configurations using the EVE-NG GUI . Follow these steps to ensure that both the topology and router configurations are retained when cloning your lab. Steps to Clone an EVE-NG Lab with Configurations Save Running Configuration on All Devices In your original lab, make sure all devices have their configurations saved to NVRAM. Go into the CLI of each router and run the command: copy running-config startup-config Export All Configurations (CFGs) On the left sidebar in the EVE-NG Web UI , click on the "More Actions" option. Then select "Export all CFGs" . This step exports the configurations of all devices in the lab. Shutdown All Devices After exporting the confi