CCNA 200-301

CCNP Enterprise

CCNP Security

CCIE Enterprise Lab

CCIE Security Lab

CCNP Service Provider

CCNP Data Center

CCNP Collaboration

CCIE DC Lab

Verifying Quality of Service (QoS) on a Cisco switch requires more than just inspecting the running configuration. Because modern Catalyst switches handle packet classification, policing, and queuing directly within hardware ASICs, a policy that looks valid in the configuration file may not actually match traffic or prevent buffer exhaustion during microbursts. Properly auditing QoS means verifying global and port-level trust boundaries, confirming that traffic matches your intended class-maps, and inspecting hardware egress queues for silent tail drops.

Before entering verification commands into the CLI, however, you must first identify which QoS architecture your Catalyst switch uses.

Legacy MLS QoS vs. Modern Cisco IOS-XE MQC Commands

A common point of confusion when checking QoS on newer switches like the Cisco Catalyst 9200, 9300, 9400, or 9500 series is running show mls qos and receiving an % Invalid input detected error.

Older Cisco IOS switches (such as the Catalyst 2960, 3560, and 3750 series) relied on Multi-Layer Switching (MLS) QoS. Under MLS QoS, prioritization was disabled by default, and enabling it globally without configuring port trust caused the switch to rewrite all incoming CoS and DSCP markings to zero. Modern Cisco IOS-XE switches (Catalyst 3650/3850 and the Catalyst 9000 family) replace MLS QoS with the Modular QoS CLI (MQC) framework and the Forwarding Engine Driver (FED). On IOS-XE platforms, QoS is permanently enabled in hardware by default, and switch ports implicitly trust incoming DSCP and CoS markings unless a service policy explicitly remarks them.

Verification Task / CriterionLegacy Cisco IOS (MLS QoS)Modern Cisco IOS-XE (MQC & FED)
Target Switch PlatformsCatalyst 2960, 2960-X/S, 3560, 3750, 3750-XCatalyst 3650, 3850, 9200, 9300, 9400, 9500
Default Global QoS StateDisabled (mls qos required to activate)Always enabled in hardware by default
Default Interface Trust StateUntrusted (rewrites CoS/DSCP to 0 when MLS is active)Implicitly trusts incoming CoS and DSCP values
Check Global QoS Statusshow mls qosActive by default; check policies via show policy-map
Check Interface Trust Stateshow mls qos interface <id> trustCheck attached policy via show run interface <id>
Verify Active Service Policiesshow policy-map interface <id>show policy-map interface <id>
Inspect Marking & Queue Mapsshow mls qos maps dscp-output-qshow table-map and show platform hardware fed switch active qos queue config interface <id>
Check Hardware Queue Dropsshow mls qos interface <id> statisticsshow platform hardware fed switch active qos queue stats interface <id>
View DSCP/CoS Traffic Countersshow mls qos interface <id> statisticsshow platform hardware fed switch active qos dscp-cos counters interface <id>

Key Takeaway: Always match your verification syntax to your switch’s operating system—use show policy-map and show platform hardware fed commands on IOS-XE Catalyst 9000 switches, and reserve show mls qos for legacy IOS platforms.

Once you know which architecture your switch runs, you can execute the exact CLI verification commands to inspect your configuration and port states.

CLI Command Template for Verifying Cisco Switch QoS Status

Use the following CLI commands to verify global QoS definitions, active interface service policies, AutoQoS profiles, and port trust configurations. Replace <interface-id> with your target switch port (for example, GigabitEthernet1/0/1 or TenGigabitEthernet1/1/1).

! ==========================================================
! 1. VERIFY GLOBAL QOS DEFINITIONS (MQC - IOS & IOS-XE)
! ==========================================================
! List all configured traffic classification rules (DSCP, CoS, ACLs)
show class-map

! Display all defined QoS policies, bandwidth allocations, and policers
show policy-map

! Verify marking translation tables (e.g., CoS-to-DSCP or policed-dscp maps)
show table-map

! ==========================================================
! 2. CHECK INTERFACE SERVICE POLICIES & PORT CONFIGURATION
! ==========================================================
! Check which service-policy or trust commands are applied to a port
show running-config interface <interface-id>

! View active ingress and egress policy statistics on a specific port
show policy-map interface <interface-id>

! Filter policy verification to only ingress (marking/policing) or egress (queuing)
show policy-map interface <interface-id> input
show policy-map interface <interface-id> output

! ==========================================================
! 3. VERIFY AUTOQOS STATUS (VOIP, VIDEO, & TRUST PROFILES)
! ==========================================================
! Display all interfaces where AutoQoS is currently enabled
show auto qos

! Check the specific AutoQoS profile (e.g., trust, voip cisco-phone) on one port
show auto qos interface <interface-id>

! ==========================================================
! 4. CHECK LEGACY MLS QOS STATUS & TRUST (CATALYST 2960/3750)
! ==========================================================
! Confirm whether MLS QoS is globally enabled on legacy IOS switches
show mls qos

! Verify port trust state (trust dscp, trust cos, untrusted) and default CoS
show mls qos interface <interface-id> trust

! Check legacy Shaped/Weighted Round Robin (SRR/WRR) queue settings
show mls qos interface <interface-id> queueing

Key Takeaway: Confirming that a policy-map exists globally is not enough; always verify that service-policy input or service-policy output is actively bound to the target <interface-id>.

Running these commands generates detailed real-time telemetry, making it essential to understand how to parse the statistical output of an active service policy.

How to Read and Interpret show policy-map interface Output

The show policy-map interface <interface-id> command is the primary tool for verifying how a Cisco switch classifies, remarks, polices, and queues live traffic. Consider the following output from an egress policy on a Catalyst switch:

Switch# show policy-map interface GigabitEthernet1/0/1 output
 GigabitEthernet1/0/1

  Service-policy output: PMAP-CAMPUS-EGRESS

    Class-map: CMAP-VOICE-PRIORITY (match-any)
      148290 packets, 29658000 bytes
      5 minute offered rate 64000 bps, drop rate 0000 bps
      Match: ip dscp ef (46)
      Priority: 10% (100000 kbps), burst bytes 2500000,
      Priority Level: 1
      Queueing
      queue limit 1200 packets
      (queue depth/total drops/no-buffer drops) 0/0/0
      (pkts output/bytes output) 148290/29658000

    Class-map: CMAP-MULTIMEDIA (match-any)
      52100 packets, 62520000 bytes
      5 minute offered rate 412000 bps, drop rate 0000 bps
      Match: ip dscp af41 (34) af42 (36)
      Queueing
      bandwidth remaining ratio 40
      queue-limit 2400 packets
      (queue depth/total drops/no-buffer drops) 0/142/0
      (pkts output/bytes output) 51958/62349600

    Class-map: class-default (match-any)
      9841220 packets, 11809464000 bytes
      5 minute offered rate 84500000 bps, drop rate 12000 bps
      Match: any
      Queueing
      bandwidth remaining ratio 25

Follow this step-by-step sequence to interpret the metrics accurately:

  1. Verify Policy Direction and Evaluation Order: Check the Service-policy input: or Service-policy output: header first. Cisco switches evaluate packets against Class-map blocks sequentially from top to bottom. Note whether the class uses match-any (logical OR) or match-all (logical AND). Any traffic that fails to match user-defined classes falls into class-default at the bottom.
  2. Validate Classification via Packet and Byte Counters: Inspect the <X> packets, <Y> bytes line and the 5 minute offered rate directly beneath each class-map name. If you have active VoIP calls on the link but CMAP-VOICE-PRIORITY shows 0 packets while class-default increments rapidly, upstream devices or ingress policies are failing to mark traffic as DSCP ef (46) (or the ingress port is rewriting markings to 0).
  3. Evaluate Priority Servicing vs. Bandwidth Ratios: Look for Priority: <X>% (<Y> kbps) and Priority Level: 1 (or 2). On Cisco switches, strict priority queues are serviced immediately ahead of standard queues to minimize latency and jitter, but they are also internally policed to their configured percentage during congestion so they cannot starve lower-priority traffic. Non-priority classes use bandwidth remaining ratio to divide the leftover link capacity proportionally.
  4. Compare Offered Packets Against Output and Drop Counters: Compare the top-level matched packets (52100 packets in CMAP-MULTIMEDIA) against (pkts output/bytes output) (51958 packets) and (queue depth/total drops/no-buffer drops) (0/142/0). Any discrepancy between matched packets and output packets is reflected in total drops or policer exceeded/violated counters (on ingress policies), pinpointing exactly which application class is experiencing congestion.

Key Takeaway: Non-zero match counters prove your classification rules work, while the difference between matched packets and (pkts output) reveals whether a specific traffic class is being throttled or dropped.

While software-polled policy-map counters provide a solid overview of class performance, diagnosing transient microbursts or buffer starvation requires inspecting the switch’s hardware ASIC queues directly.

How to Check and Troubleshoot QoS Queue Drops at the Hardware Level

When users report voice choppiness, video pixelation, or TCP retransmissions—especially on ports stepping down from 10 Gbps uplinks to 1 Gbps access links—hardware egress buffers may be overflowing even when average link utilization appears low. Follow these steps to isolate and troubleshoot hardware queue drops:

Step 1: Check High-Level Interface Output Drops

Start by checking whether the physical interface reports aggregate outbound drops:

Switch# show interfaces GigabitEthernet1/0/1 | include output drops|Queueing
  Input queue: 0/2000/0/0 (size/max/drops/flushes); Total output drops: 48291
  Queueing strategy: Class-based queueing

If Total output drops is incrementing, one or more hardware egress queues on that port are exhausting their allocated buffer thresholds.

Step 2: Inspect Hardware ASIC Queue Statistics (IOS-XE / Catalyst 9000)

On Catalyst 3650/3850 and Catalyst 9000 switches, query the Forwarding Engine Driver (FED) to see exact enqueue and drop counts per hardware queue (Queue 0 through Queue 7) and per drop threshold (TH0, TH1, TH2):

Switch# show platform hardware fed switch active qos queue stats interface GigabitEthernet1/0/1
----------------------------------------------------------------------------------------------
Asic:0 Core:0 DATA Port:0 GPN:1 LinkSpeed:0x1
Interface: GigabitEthernet1/0/1 (0x1)
----------------------------------------------------------------------------------------------
Queue   Enqueue-TH0         Enqueue-TH1         Enqueue-TH2         
----------------------------------------------------------------------------------------------
0       0                   0                   4589210             
1       148290              0                   0                   
2       0                   51958               0                   
----------------------------------------------------------------------------------------------
Queue   Drop-TH0            Drop-TH1            Drop-TH2            SBufDrop            QebDrop
----------------------------------------------------------------------------------------------
0       0                   0                   48149               48149               0
1       0                   0                   0                   0                   0
2       0                   142                 0                   142                 0

Interpret the hardware FED fields as follows:
* Enqueue-TH0 / TH1 / TH2: Packets successfully admitted into the hardware queue under Threshold 0, Threshold 1, or Threshold 2 (higher thresholds typically accommodate higher-priority DSCP markings within the same queue).
* Drop-TH0 / TH1 / TH2: Weighted Tail Drop (WTD) counters showing packets discarded because that specific threshold exceeded its limit.
* SBufDrop (Shared Buffer Drops): Packets dropped because the queue consumed its dedicated buffer reservation and the ASIC’s shared memory pool was also full—a classic signature of microburst congestion.
* QebDrop (Queue Entry Buffer Drops): Packet descriptor exhaustion at the ASIC level, usually seen during extreme small-packet floods.

Step 3: Check Hardware Queue Drops on Legacy MLS QoS Switches

If you are troubleshooting a legacy Catalyst 2960 or 3750 switch (as differentiated in the platform table earlier), run show mls qos interface <interface-id> statistics and inspect the bottom output queues dropped block:

  output queues enqueued:
   queue:    threshold1   threshold2   threshold3
  -----------------------------------------------
   queue 0:           0         1420        89210
   queue 1:      541200            0            0

  output queues dropped:
   queue:    threshold1   threshold2   threshold3
  -----------------------------------------------
   queue 0:           0           84         3102
   queue 1:           0            0            0

Step 4: Baseline Counters and Tune Queue Buffers

Because hardware counters accumulate since the last reboot, clear the statistics to verify whether drops are actively occurring right now:

! On Cisco IOS-XE switches
Switch# clear platform hardware fed switch active qos statistics interface GigabitEthernet1/0/1

! Standard interface counter reset
Switch# clear counters GigabitEthernet1/0/1

If non-priority queues (Queue 0 or class-default) show recurring SBufDrop increments during normal business hours while average link bandwidth remains well below 100%, increase the soft buffer allocation inside your egress policy-map using the queue-buffers ratio command on IOS-XE (or mls qos queue-set output <id> buffers on legacy IOS).

Key Takeaway: Always inspect hardware-level FED queue counters (Drop-THx and SBufDrop) to distinguish harmless bulk-traffic tail drops in Queue 0 from critical real-time packet loss in priority queues.

Because CLI counters only reflect cumulative totals since the last reset, tracking intermittent congestion over days or weeks requires polling these QoS metrics into a centralized monitoring platform.

Best Network Monitoring Tools to Track Cisco Switch QoS Metrics

Relying solely on CLI commands makes it difficult to correlate temporary voice or application degradation with overnight backup spikes or daytime microbursts. Network monitoring tools solve this by polling Cisco’s CISCO-CLASS-BASED-QOS-MIB (CBQoS MIB) via SNMP or subscribing to IOS-XE Model-Driven Telemetry streams.

  • 1. Cisco Catalyst Center (formerly Cisco DNA Center): Cisco’s native management and assurance appliance for Catalyst 9000 environments. It uses streaming telemetry and SNMP to visualize per-queue bandwidth utilization, end-to-end DSCP marking preservation, and hardware queue drops across the fabric without manual MIB configuration.
  • 2. SolarWinds Network Traffic Analyzer (NTA) & NPM: Combines NetFlow/IPFIX flow analysis with SNMP CBQoS polling. SolarWinds charts pre-policy versus post-policy byte rates, highlights policer drops per class-map, and lets engineers correlate DSCP class spikes with specific top-talker IP addresses.
  • 3. ManageEngine OpManager (with NetFlow Analyzer Add-on): Offers built-in Cisco CBQoS monitoring templates that break down traffic consumption by class-map, queuing policy, and match statement. It supports automated threshold alerts whenever egress queue drop rates exceed defined percentages.
  • 4. Paessler PRTG Network Monitor: Provides a dedicated SNMP Cisco CBQoS Sensor that tracks individual class-map throughput, pre-policy/post-policy bytes, buffer drops, and policer violations with granular per-channel alerting.
  • 5. Telegraf + Prometheus + Grafana (Open-Source Telemetry Stack): Ideal for modern Cisco IOS-XE switches supporting Model-Driven Telemetry (gNMI/NETCONF). By streaming YANG operational models (such as Cisco-IOS-XE-interfaces-oper and policy-map statistics) at 5- to 10-second intervals, this stack detects sub-minute microburst drops that standard 5-minute SNMP pollers miss.

Key Takeaway: Use SNMP CBQoS polling for historical trend reporting across mixed switch fleets, or deploy streaming telemetry via Catalyst Center or Telegraf/Grafana to catch sub-minute microburst drops on Catalyst 9000 switches.

When configuring telemetry collectors or auditing platform-specific hardware queue limits, always cross-reference Cisco’s authoritative platform documentation.

Official Cisco Catalyst QoS Documentation and Command References

Hardware queue architectures, buffer ratios, and supported MQC keywords vary by ASIC generation (such as UADP 2.0, UADP 3.0, and Cisco Silicon One) and IOS-XE release train. Reference the following official Cisco documentation manuals and MIB specifications for platform-exact verification definitions:

  • Quality of Service Configuration Guide, Cisco IOS XE (Catalyst 9200 / 9300 / 9400 / 9500 Switches): The primary platform specification grouped by IOS-XE release (16.x and 17.x). Defines default 8-queue (2P6Q3T) hardware mappings, MQC restrictions, AutoQoS macro expansions, and FED verification commands.
  • Cisco IOS Quality of Service Solutions Command Reference: The alphabetical syntax dictionary detailing every command argument and output field for show policy-map interface, show class-map, show table-map, and show auto qos.
  • Catalyst 3750-X and 3560-X / Catalyst 2960-X Switch Software Configuration Guide (Configuring QoS): The authoritative engineering manual for legacy MLS QoS platforms, specifying Shaped Round Robin (SRR) weight calculations, queue-set buffer allocations, and show mls qos diagnostic syntax.
  • Enterprise QoS Solution Reference Network Design (SRND) & Cisco Validated Designs (CVD): Official architecture guides detailing RFC 4594 DSCP marking standards, campus trust boundary placement, and recommended queue buffer ratios for voice, video, and transactional data.
  • Cisco SNMP Object Navigator — CISCO-CLASS-BASED-QOS-MIB (OID 1.3.6.1.4.1.9.9.166): The formal MIB specification mapping CLI show policy-map interface counters to SNMP tables, including cbQosCMStatsTable (class-map traffic counters), cbQosQueueingStatsTable (queue depths and drops), and cbQosPoliceStatsTable (conformed, exceeded, and violated policer statistics).

Key Takeaway: Consult the platform-specific Quality of Service Configuration Guide matching your exact switch model and IOS-XE version whenever verifying ASIC queue-to-threshold mappings or hardware buffer limits.

Checking QoS on a Cisco switch follows a clear hierarchy: confirm your switch platform’s QoS architecture (IOS-XE MQC versus legacy MLS QoS), verify that service policies are bound to the right interfaces, confirm live traffic classification in show policy-map interface, and inspect hardware ASIC counters for threshold or shared buffer drops. Combining these targeted CLI checks with continuous CBQoS or streaming telemetry monitoring ensures your critical voice, video, and business traffic remains protected even during peak network congestion.

Please follow and like us:
Last modified: November 17, 2026

Author

Comments

Write a Reply or Comment

Your email address will not be published.