CCNA 200-301

CCNP Enterprise

CCNP Security

CCIE Enterprise Lab

CCIE Security Lab

CCNP Service Provider

CCNP Data Center

CCNP Collaboration

CCIE DC Lab

Maximum Transmission Unit (MTU) defines the maximum size of a packet or frame—measured in bytes—that a network interface can process and forward without fragmenting it. In enterprise and data center networks, adjusting the MTU beyond the standard Ethernet size of 1500 bytes to accommodate jumbo frames (typically up to 9000 or 9216 bytes) is essential for optimizing high-throughput traffic, such as iSCSI storage, NFS shares, hypervisor migrations, and network virtualization overlays like VXLAN and GRE.

Configuring MTU incorrectly or allowing mismatches across adjacent network links introduces dropped frames, unexplained throughput degradation, and routing protocol neighbor failures. This comprehensive guide covers how to configure, automate, verify, and troubleshoot interface MTU across modern and legacy Cisco switch architectures.


Understanding System MTU vs. Interface MTU

What is the difference between System MTU and Interface MTU?

System MTU is a global setting that establishes the baseline default MTU for all interfaces across the entire switch. When you modify the system MTU, any port without an explicit local override inherits this global value.

Interface MTU (also referred to as Per-Port MTU) is applied directly within an individual interface configuration mode—such as a physical Ethernet port, an EtherChannel (Port-Channel), or a Switched Virtual Interface (SVI). Configuring an MTU at the interface level overrides the global system baseline for that specific port.

Why did older Cisco Catalyst switches require a reload to change MTU?

On legacy Cisco Catalyst platforms (such as the Catalyst 2960, 3560, and 3750 series running classic Cisco IOS), switch ASIC packet buffers were statically allocated during initialization. Changing the maximum frame size with commands like system mtu jumbo <bytes> required a full hardware reload (reload) to repartition memory buffers.

In contrast, modern Catalyst switches (including the Catalyst 3650, 3850, and the entire Catalyst 9000 series running Cisco IOS-XE) support dynamic ASIC buffer management. As of Cisco IOS-XE Amsterdam 17.1.1, these platforms support per-port MTU adjustments that take effect immediately without requiring a reboot.

What is the default MTU on Cisco switch interfaces?

The default MTU on all Cisco Catalyst switch interfaces is 1500 bytes. This value represents the maximum Layer 3 payload size (the IP packet, including IP and transport headers) that standard Ethernet frames carry. The Layer 2 frame overhead (a 14-byte Ethernet header plus a 4-byte Frame Check Sequence, along with any optional 4-byte 802.1Q VLAN tags) is tracked separately by the switch hardware.

How do System MTU and Interface MTU interact?

When an interface is configured with a specific per-port MTU, that port uses its local value regardless of the system MTU. If you remove the port-level configuration using the no mtu command, the interface automatically reverts to the global system MTU value. On some platforms, the global system MTU acts as an absolute hardware ceiling; attempting to set an interface MTU higher than the supported system limit will cause the switch CLI to reject the command.


Step-by-Step Guide: Configuring Interface MTU on Cisco Catalyst (IOS-XE)

Modern Cisco Catalyst 9000 switches running Cisco IOS-XE allow you to configure MTU on physical switchports (Layer 2), routed ports (Layer 3), Port-Channels, and SVIs. Follow these steps to apply interface-specific MTU settings.

Step 1: Enter Global Configuration Mode

Log in to the switch and enter privileged EXEC mode, then move to global configuration:

Switch# configure terminal
Switch(config)#

Step 2: Configure MTU on a Layer 2 Switchport (Access or Trunk)

To configure MTU on a switched interface (either access or trunk mode), apply the mtu command directly under the physical interface:

Switch(config)# interface GigabitEthernet 1/0/1
Switch(config-if)# description Storage-Host-Trunk
Switch(config-if)# switchport mode trunk
Switch(config-if)# mtu 9000
Switch(config-if)# exit

Note: On IOS-XE 17.1.1 and later, changing the MTU on a standalone physical port does not cause the link to flap.

Step 3: Configure MTU on a Layer 3 Routed Port

For routed physical interfaces (configured with no switchport), you can configure the interface MTU as well as protocol-specific MTUs (such as ip mtu or ipv6 mtu):

Switch(config)# interface GigabitEthernet 1/0/2
Switch(config-if)# description Routed-Uplink-to-Core
Switch(config-if)# no switchport
Switch(config-if)# ip address 10.10.10.1 255.255.255.0
Switch(config-if)# mtu 9000
Switch(config-if)# ip mtu 9000
Switch(config-if)# exit

Note: If both mtu and ip mtu are configured, ip mtu sets the IP packet fragmentation boundary, which cannot exceed the physical mtu value.

Step 4: Configure MTU on a Switched Virtual Interface (SVI)

To support jumbo frames routed across a VLAN, update the MTU on the corresponding SVI (VLAN interface):

Switch(config)# interface Vlan 100
Switch(config-if)# description iSCSI-VLAN-Gateway
Switch(config-if)# ip address 192.168.100.1 255.255.255.0
Switch(config-if)# mtu 9000
Switch(config-if)# ip mtu 9000
Switch(config-if)# exit

Note: Increasing the MTU on an SVI will not pass jumbo frames unless all underlying physical Layer 2 switchports carrying that VLAN are also configured to support that MTU size.

Step 5: Configure MTU on a Port-Channel (EtherChannel)

When configuring an EtherChannel, apply the MTU setting directly to the logical Port-channel interface:

Switch(config)# interface Port-channel 1
Switch(config-if)# mtu 9000
Switch(config-if)# exit

IOS-XE automatically propagates the MTU setting from the logical Port-Channel interface to all active physical bundle members. If the bundle runs dynamic LACP or PAgP, the port-channel interface will momentarily flap during the MTU update to negotiate new parameters.

Step 6: Save the Configuration

Exit configuration mode and write the active running configuration to NVRAM:

Switch(config)# end
Switch# write memory

CLI Configuration Templates for Bulk MTU Changes

When provisioning a switch for hypervisor clusters, SAN environments, or core uplinks, configuring interfaces individually is inefficient. Use these ready-to-deploy templates to apply MTU settings across multiple interfaces simultaneously.

Template 1: Bulk Physical Access/Trunk Ports Using Interface Range

Applies a 9000-byte jumbo MTU to an entire block of physical ports:

! --- Apply Jumbo MTU across contiguous ports ---
configure terminal
interface range GigabitEthernet 1/0/1 - 24
 description Hypervisor-Host-Facing-Ports
 mtu 9000
exit
end
write memory

Template 2: Core Infrastructure (Uplink Trunks, Port-Channels, and SVIs)

Sets jumbo MTU across multiple uplinks, aggregated links, and routed VLAN interfaces:

! --- Bulk Core Uplink and SVI MTU Template ---
configure terminal

! Apply to multiple Port-Channels
interface range Port-channel 1 - 4
 mtu 9216
exit

! Apply to routed VLAN interfaces
interface range Vlan 10, Vlan 20, Vlan 30, Vlan 40
 mtu 9000
 ip mtu 9000
exit

end
write memory

Template 3: Global System Baseline MTU Configuration

Sets the global default MTU to jumbo across all interfaces on the switch, removing the need for per-port configurations on devices where all traffic should default to jumbo frames:

! --- Global Switch System MTU Configuration ---
configure terminal
system mtu 9000
end
write memory

Verifying MTU Settings and Testing Jumbo Frames

After applying MTU changes, verify that the switch hardware and the operational control plane reflect the new settings, and perform end-to-end packet validation.

Step 1: Verify Configured MTU on Interfaces

On Cisco IOS-XE switches, verify interface MTUs using show interface mtu or by targeting specific ports:

Switch# show interface mtu

Port         Name                     MTU
Gi1/0/1      Storage-Host-Trunk       9000
Gi1/0/2      Routed-Uplink-to-Core    9000
Po1          Core-Interconnect        9216
Vl100        iSCSI-VLAN-Gateway       9000

To view the complete operational status of a single port, including Layer 2 and IP MTUs, filter the interface output:

Switch# show interfaces GigabitEthernet 1/0/1 | include MTU
  MTU 9000 bytes, BW 1000000 Kbit/sec, DLY 10 usec, reliability 255/255, txload 1/255, rxload 1/255

Switch# show ip interface GigabitEthernet 1/0/2 | include MTU
  MTU is 9000 bytes

Step 2: Validate End-to-End MTU with Do-Not-Fragment (DF) Pings

A show command only validates local switch configuration; it does not guarantee that every intermediate hop, firewall, or remote endpoint supports that frame size. To test end-to-end MTU capacity, execute an ICMP ping with the Do Not Fragment (DF) bit set.

Testing from Cisco IOS-XE CLI

In Cisco IOS and IOS-XE, the size argument defines the total IP packet size (IP header + ICMP header + payload):

Switch# ping 10.10.10.2 size 9000 df-bit
Type escape sequence to abort.
Sending 5, 9000-byte ICMP Echos to 10.10.10.2, timeout is 2 seconds:
Packet sent with the /Dont Fragment/ bit set
!!!!!
Success rate is 100 percent (5/5), round-trip min/avg/max = 1/2/3 ms

If any link along the path cannot accommodate 9000 bytes, the switch output displays M (fragmentation needed and DF set) or drops:

Switch# ping 10.10.10.2 size 9000 df-bit
Sending 5, 9000-byte ICMP Echos to 10.10.10.2, timeout is 2 seconds:
Packet sent with the /Dont Fragment/ bit set
.....
Success rate is 0 percent (0/5)

Testing from Connected Hosts (Windows and Linux)

Unlike Cisco CLI, host operating system ping tools specify only the ICMP payload size, excluding the 20-byte IP header and 8-byte ICMP header (28 bytes total). To test for a 9000-byte MTU, send a payload of 8972 bytes (8972 + 28 = 9000):

  • Microsoft Windows:
    cmd
    ping -f -l 8972 10.10.10.2

    (Parameter -f sets the DF bit; -l specifies buffer length.)

  • Linux:
    bash
    ping -M do -s 8972 10.10.10.2

    (Parameter -M do prohibits fragmentation; -s sets payload size.)


Troubleshooting MTU Mismatches and OSPF EXSTART Hangs

Mismatched MTUs are among the most common root causes of routing instability and silent packet loss. When two directly connected devices do not share identical MTU values, critical control-plane and data-plane failures occur.

Issue 1: OSPF Adjacency Stuck in EXSTART / EXCHANGE

When forming an adjacency over an interface, OSPF routers transition through EXSTART and EXCHANGE states, where they determine master/slave roles and exchange Database Description (DBD) packets. Under RFC 2328, each router advertises its local interface MTU in the DBD packet.

Router-A (MTU 9000) <--------------------------> Router-B (MTU 1500)
                    --- DBD (MTU: 9000) --->
                    <--- DBD (MTU: 1500) ---

If Router A has an MTU of 9000 and Router B has an MTU of 1500:
1. Router B receives Router A’s DBD packet advertising an MTU of 9000. Because 9000 exceeds Router B’s local MTU (1500), Router B rejects the packet and refuses to proceed.
2. Router A keeps retransmitting its DBD packet because Router B never acknowledges it.
3. The neighbor state remains permanently trapped in EXSTART/EXCHANGE.

Step-by-Step Resolution

  1. Verify the neighbor state and identify the stuck interface:
    text
    Switch# show ip ospf neighbor
    Neighbor ID Pri State Dead Time Address Interface
    192.168.1.2 1 EXSTART/DR 00:00:34 10.10.10.2 GigabitEthernet1/0/2
  2. Enable OSPF adjacency debugging to confirm the MTU mismatch:
    text
    Switch# debug ip ospf adj
    OSPF: Rcv DBD from 192.168.1.2 on GigabitEthernet1/0/2 with larger MTU 9000 (local MTU 1500)
    OSPF: Mismatched MTU, DBD packet dropped
  3. Correct the MTU on the mismatched interface so that both peers align:
    text
    Switch(config)# interface GigabitEthernet 1/0/2
    Switch(config-if)# mtu 9000
    Switch(config-if)# ip mtu 9000
  4. Temporary Emergency Workaround: If you cannot immediately change the MTU on one side due to administrative constraints, disable OSPF MTU negotiation on the interface:
    text
    Switch(config-if)# ip ospf mtu-ignore

    Caution: ip ospf mtu-ignore should only be used as a temporary workaround. While it allows OSPF to achieve a FULL state, any subsequent Link State Update (LSU) packet larger than 1500 bytes will be silently dropped by the lower-MTU peer, causing inconsistent routing tables.

Issue 2: Data-Plane Packet Drops and “Giant” Counters

When an upstream switch or server transmits a 9000-byte frame into a switchport configured for 1500 bytes, the ingress switch hardware discards the frame as an invalid size.

Diagnostic Steps:

  1. Check the physical interface counters for dropped packets and errors:
    text
    Switch# show interfaces GigabitEthernet 1/0/1 | include giants|input errors|CRC
    14285 input errors, 14285 giants, 0 CRC, 0 frame
  2. A non-zero, continuously incrementing giants counter confirms that incoming frames exceed the configured interface MTU.
  3. Resolve the drop condition by raising the interface MTU to match the transmitting device:
    text
    Switch(config)# interface GigabitEthernet 1/0/1
    Switch(config-if)# mtu 9216

Cisco Catalyst vs. Nexus (NX-OS) MTU Configuration Differences

While Catalyst switches handle MTU via straightforward interface commands or global system defaults, Cisco Nexus data center switches running NX-OS use different architectural workflows depending on the hardware platform.

Architectural FeatureCisco Catalyst (IOS / IOS-XE)Cisco Nexus (NX-OS)
Primary OSCisco IOS-XE (Catalyst 9000 / 3850)Cisco NX-OS (Nexus 3000 / 5000 / 7000 / 9000)
Default L2 / L3 MTU1500 bytes1500 bytes
Maximum Supported JumboUp to 9216 bytes (Platform-dependent)Up to 9216 bytes
Layer 2 MTU ConfigurationPer-port mtu <bytes> directly under interfaceSystem-wide via policy-map type network-qos (N5k/N6k/older N3k) OR per-port mtu 9216 (N7k/N9k)
Layer 3 MTU Configurationmtu and ip mtu under routed port / SVImtu <bytes> under routed port / SVI
Global Baseline Commandsystem mtu <bytes>system jumbomtu <bytes>
Reboot Required?No on modern IOS-XE; Yes on legacy classic IOSNever requires a reboot
Verification Commandshow interface mtu
show interfaces <id> \| inc MTU
show interface <id> \| inc MTU
show queuing interface <id>

The NX-OS Network-QoS Difference

On traditional Nexus switches (such as Nexus 5000, 5500, 6000, and fixed 3000 series), Layer 2 jumbo frames cannot be configured on individual interfaces. Instead, the administrator must configure a global Quality of Service policy map:

! --- Nexus 5000/6000 Global QoS Jumbo MTU Configuration ---
Nexus-5K(config)# policy-map type network-qos jumbo
Nexus-5K(config-pmap-nq)# class type network-qos class-default
Nexus-5K(config-pmap-c-nq)# mtu 9216
Nexus-5K(config-pmap-c-nq)# exit
Nexus-5K(config-pmap-nq)# exit
Nexus-5K(config)# system qos
Nexus-5K(config-sys-qos)# service-policy type network-qos jumbo

On modern Nexus 9000 switches running NX-OS in standalone mode, Cisco brought the interface behavior in line with Catalyst switches, allowing engineers to apply mtu 9216 directly to Layer 2 and Layer 3 interfaces.


Cisco Catalyst MTU Technical Reference and Command Specifications

MTU Framing and Wire Overhead

When planning an MTU strategy, differentiate between Layer 3 packet size and the total Layer 2 Ethernet frame size transmitted onto the physical medium.

+-------------------------------------------------------------------------------+
|                      Layer 2 Ethernet Frame (1518 - 9238 bytes)               |
+-------------------+---------------+-----------------------+-------------------+
|  Ethernet Header  | 802.1Q Tag    |   IP Packet (MTU)     | Frame Check (FCS) |
|     (14 Bytes)    | (4 Bytes opt) | (1500 - 9216 Bytes)   |     (4 Bytes)     |
+-------------------+---------------+-----------------------+-------------------+
  • Standard Ethernet Frame: 1500 (IP MTU) + 14 (MAC header) + 4 (FCS) = 1518 bytes.
  • Standard 802.1Q Tagged Frame: 1500 (IP MTU) + 14 (MAC header) + 4 (VLAN tag) + 4 (FCS) = 1522 bytes.
  • Jumbo Ethernet Frame (9000): 9000 (IP MTU) + 14 (MAC header) + 4 (FCS) = 9018 bytes (or 9022 bytes with 802.1Q tagging).
  • Maximum Platform Frame (9216): 9216 (IP MTU) + 14 (MAC header) + 4 (FCS) = 9234 bytes (or 9238 bytes with 802.1Q tagging).

Note: Cisco Catalyst switches define the mtu command as the Layer 3 payload size (IP MTU). The switch automatically adjusts the Layer 2 frame admission hardware to accommodate the required MAC headers and tags without manual overhead calculations.

Platform Support Specifications

  • Catalyst 9200 Series: Supports system-wide and per-port MTU up to 9216 bytes.
  • Catalyst 9300 / 9400 / 9500 / 9600 Series: Supports system-wide and per-port MTU up to 9216 bytes across all Ethernet port types.
  • Minimum IPv6 MTU: Enforced across all Catalyst 9000 switches at 1280 bytes in adherence to RFC 8200.

Command Reference

system mtu <1500-9216>

  • Mode: Global Configuration
  • Description: Configures the global default MTU applied to all interfaces that do not have an explicit per-port MTU configuration.
  • Default: 1500 bytes.

mtu <1500-9216>

  • Mode: Interface Configuration (Physical, Port-Channel, SVI)
  • Description: Sets the maximum transmission unit size for switched or routed traffic passing through the interface.
  • Default: Inherits global system mtu (1500 bytes).

ip mtu <68-9216>

  • Mode: Interface Configuration (Routed port or SVI)
  • Description: Sets the IP packet size before Layer 3 fragmentation occurs. Must be less than or equal to the interface mtu.

show interface mtu

  • Mode: Privileged EXEC
  • Description: Displays a consolidated table of all configured and operational MTUs across all switchports on the device.

Conclusion

Changing the MTU on a Cisco switch interface requires aligning your configuration method with the platform’s underlying operating system. On modern Catalyst switches running Cisco IOS-XE, per-port MTU allows seamless changes without service interruptions or reboots. For consistent performance, ensure that every device along the end-to-end data path—from host NICs to intermediate switches, routers, and firewalls—shares an identical MTU. Finally, always validate real-world forwarding using extended pings with the Do Not Fragment (DF) bit set to confirm seamless transport for standard and jumbo traffic alike.

Please follow and like us:
Last modified: November 16, 2026

Author

Comments

Write a Reply or Comment

Your email address will not be published.