Use the simulator to see how a packet moves from a sender to a server when one link has a smaller Maximum Transmission Unit (MTU) than the links around it. Change the IP version, packet size, tunnel MTU, and feedback policy to compare forwarding, fragmentation, Path MTU Discovery, and a black-hole symptom.
This is a browser-based teaching model. It does not send packets from your device or change your network settings.
MTU is a link’s limit for the network-layer packet it can carry. For IP, that means the IP header plus the IP payload. The Ethernet header and frame check sequence are outside that IP MTU calculation. A common Ethernet IP MTU is 1500 bytes, but networks, tunnels, and interfaces can use other values.
TCP Maximum Segment Size (MSS) describes the maximum amount of TCP data an endpoint is prepared to receive in one TCP segment. Endpoints can advertise an MSS option in TCP SYN segments. Each endpoint advertises its own receive limit; TCP does not negotiate one shared MSS number. The sender must also fit the segment inside the IP packet size its path can carry.
With the basic headers and a 1500-byte IP MTU:
| Example | Calculation | TCP data limit |
|---|
| IPv4 with a 20-byte IP header and 20-byte TCP header | 1500 − 20 − 20 | 1460 bytes |
| IPv6 with a 40-byte base header and 20-byte TCP header | 1500 − 40 − 20 | 1440 bytes |
These are common baseline calculations, not guaranteed values for every packet. TCP options, IPv4 options, IPv6 extension headers, and receiver limits can reduce the data that fits.
Encapsulation adds outer headers around the original packet. Encryption, tunnel protocol, padding, and the underlay path all affect the space available for the inner IP packet. That is why a packet that fits a 1500-byte Ethernet link can be too large for a tunnel path. There is no single tunnel MTU that applies to every VPN or provider network.
When you use the simulator, compare the IP packet size with the configured path MTU. The lab keeps the local link at 1500 bytes and lets you adjust one constrained path segment. The topology and values are illustrative; measure the MTU for the real path before changing a production interface.
If an IPv4 packet is too large for the next link and fragmentation is allowed, a router may split the datagram into fragments. Each fragment gets an IPv4 header; the destination host reassembles the original datagram. Fragment offsets use 8-byte units, so non-final fragment data is aligned on an 8-byte boundary.
If the IPv4 Don’t Fragment (DF) flag is set, the router drops a packet that cannot fit and can report the limit with ICMP Destination Unreachable, “Fragmentation Needed” (Type 3, Code 4). Classic IPv4 Path MTU Discovery uses this feedback so the sender can reduce packet size instead of relying on router fragmentation.
Transit routers do not fragment IPv6 packets. If a packet is larger than the next link’s MTU, a router drops it and can send ICMPv6 “Packet Too Big” (Type 2, Code 0) back to the source. The source then sends smaller packets; if IPv6 fragmentation is needed, it is performed by the source using a Fragment extension header.
If the required ICMP message cannot reach the sender, Path MTU Discovery may not learn about the smaller link. Small requests can work while larger transfers stall or time out. This symptom is often called a Path MTU black hole. It is a clue to investigate, not proof of a particular firewall fault: compare packet sizes and check both directions of the path.
Packetization-Layer Path MTU Discovery (PLPMTUD) is another approach. It probes packet sizes at a higher layer and can reduce dependence on ICMP error feedback. It still requires protocol and endpoint support. RFC 4821 describes a general IP PLPMTUD method; RFC 8899 specifies PLPMTUD for datagram transports.
- Confirm the source and destination, IP version, and whether the problem affects one direction or both.
- Check interface MTUs along the path, including tunnel interfaces and encapsulation overhead.
- Compare small and larger probes. A small successful ping shows that the small probe can pass; it does not prove larger application packets will pass.
- Check that necessary ICMP errors are allowed back to the source. Avoid a blanket “block all ICMP” rule.
- If the network design uses TCP MSS adjustment, verify the SYN/SYN-ACK MSS values and the effective packet size at the tunnel boundary.
- Change one setting at a time, then repeat the same test from both ends where possible.
For a controlled IPv4 path test on Linux, ping -4 -M do -s 1472 <destination> sends 1472 bytes of ICMP data. With the 8-byte ICMP header and 20-byte IPv4 header, that is a 1500-byte IP packet, assuming no IPv4 options. Windows has the similar form ping -f -l 1472 <destination>. Choose a destination you are authorized to test, and adjust the probe size to the path you are measuring. The command’s payload size is not itself the MTU.
tracepath <destination> on Linux can also report path MTU information when supported by the route and the network. Tool output and behavior vary by operating system and network policy.
- MTU limits an IP packet on a link; MSS limits TCP data in one segment.
- Headers consume part of the packet budget. Ethernet framing is not part of the IP MTU value.
- A tunnel can lower the effective path MTU because of encapsulation overhead.
- IPv4 may fragment when permitted; IPv6 routers do not fragment packets in transit.
- PMTUD feedback must be able to reach the sender, or larger traffic may stall.
- Use measured path values and the relevant platform documentation before making network changes.
QDoes a 1500-byte MTU mean 1500 bytes of application data?+
ANo. The MTU applies to the IP packet, including the IP header. TCP and other transport headers also use space, so the application data limit is smaller.
QIs MSS the same as MTU?+
ANo. MTU is an IP packet limit on a link. MSS is a TCP data limit advertised by a receiver; the sender also has to respect the IP packet size allowed by its path.
QWhy can ping work while a website or file transfer stalls?+
AA small ping may fit even when larger packets do not. If a smaller path MTU is present and the sender does not receive the required error feedback, larger TCP traffic can stall. Ping is one diagnostic clue, not a complete application test.
QDo IPv6 routers fragment oversized packets?+
ANo. A transit router drops an oversized IPv6 packet and can send an ICMPv6 Packet Too Big message. If fragmentation is needed, the source host performs it.
Continue with the Internet Fundamentals topic or explore why ping can work while an application fails. You can also use the subnet calculator when checking address ranges alongside network sizing.
Pass it along
Found this useful? Share it.
Choose a platform or copy the direct link.