Maximum Transmission Unit

What is MTU?

Maximum transmission unit, or MTU, is the largest data packet a network device will send as a single unit. It is measured in bytes.

Networks don't move data in one lump. They break it into packets first, so a file, an email, or a video stream all travel as a series of small pieces.

How large any single piece can get is exactly what MTU governs. On most networks that limit lands at 1,500 bytes. Go past that on a link that cannot handle it and the packet will not make it through whole. The network either splits it up or discards it.

Here is why anyone running monitoring should care. MTU tends to sit quietly in the background until the day it does not, and when it goes wrong, it rarely looks like an MTU problem. You see slow links. You see packets dropping. You see a connection that just fails, with nothing in the logs pointing at the real cause.

Set it correctly and traffic moves the way it should. Get it wrong, though, and you inherit a whole class of slowdowns that are genuinely hard to pin down.

How Does MTU Work?

MTU is a limit you set on a network interface. Every packet leaving that interface has to fit within it, no exceptions.

A packet has two parts worth knowing about. One part, the header, holds addressing information. What's left, the payload, is the data itself. MTU measures both together, so the number covers the entire packet and not just the useful part inside. It applies to Internet Protocol packets at the network layer.

There is one detail that catches people out. Packets rarely move straight from one device to the next. They hop across routers and switches on the way, and each of those keeps its own MTU. The lowest value on the route is the one that matters. Whatever it is, it caps the whole trip.

Why Is the Standard MTU 1,500 Bytes?

That 1,500-byte number is old. It traces back to early Ethernet and was never really changed after. Networks and devices still treat it as the default today. A full Ethernet frame runs to 1,518 bytes. Strip out the small header and trailer that handle addressing, and 1,500 bytes remain for the payload. That payload is the IP packet, so 1,500 turned into the figure everything else was built around.

Reach for almost any device today and it hands you this value by default. It lands in a sensible spot, efficient enough to move real data, standard enough to work with everything else on the network.

What Happens When a Packet Exceeds the MTU?

So what happens when a packet is simply too big for the next device in line? One of two things, and neither helps performance.

1. Fragmentation

The first is fragmentation. Here the router takes the oversized packet, divides it into pieces small enough to pass, and the receiver rebuilds the original at the far end. This works. It also adds overhead and a little delay, and it carries a risk: lose a single fragment in transit and the entire packet usually has to be sent again.

2. Dropped packets

The second is a straight drop. Certain packets cannot be divided at all. When one of them meets a limit it exceeds, the router drops it and moves on. IPv6 packets are the usual example. IPv6 does not allow routers to fragment packets partway through a route, so anything oversized is dropped along the way.

3. The Don't Fragment flag

A packet can also force the issue. With the Don't Fragment flag set in its header, every router on the path is barred from splitting it. Say that packet then reaches a link too small to carry it. The router drops it and sends an error back to the sender. Left unnoticed, this becomes one of networking's more maddening faults: small requests pass without trouble, then the connection stalls the second someone moves a large file.

What Are Jumbo Frames?

Jumbo frames go beyond the standard 1,500 bytes, reaching as high as 9,000. The appeal is simple arithmetic. Bigger packets mean fewer of them for the same amount of data, which cuts overhead and lightens the processing load on fast links. Throughput rises as a result.

You will find jumbo frames mostly in data centers and storage networks, where traffic is both heavy and predictable. There is a condition attached, though. Every device on the path has to support the larger size. Miss one, and you are straight back to fragmentation and dropped packets.

How Is MTU Different From MSS?

MTU and MSS both put a cap on size, but they are not measuring the same thing. MTU is the whole packet, headers and all. MSS, or maximum segment size, is narrower. It counts only the payload inside a TCP segment and sets the headers aside.

The link between the two is plain arithmetic: MSS generally works out to MTU minus 40 bytes, the room those standard IP and TCP headers occupy.

What Is Path MTU Discovery?

Path MTU Discovery, or PMTUD, lets a sender find the smallest MTU on a route in advance, before it settles on a packet size. Heading off fragmentation from the very start is the whole aim. It does this by sending test packets marked so they cannot be split.

When a device along the way cannot handle the size, it replies with an error that names its own limit. The sender takes that in, reduces its packet size, and tries again. It repeats the cycle until packets finally cross the full path without being dropped.

Why Does MTU Matter for Network Performance?

MTU has a quiet influence on how well a network moves traffic. Matched to the path, packets flow cleanly, with no fragmentation to slow them. The trouble starts at the edges. An MTU set too high causes drops on links that cannot cope with the size, while one set too low wastes network bandwidth, since it now takes more packets to carry the same load.

This is the point where MTU and monitoring meet. A mismatch almost never declares itself openly. Instead it shows up secondhand, as higher latency, as packet loss isolated to one link, or as an application that feels sluggish for no reason anyone can immediately name.

So when a network issue refuses to add up, MTU is worth ruling out early. Keeping it consistent across connected devices remains one of the least glamorous and most dependable ways to hold performance steady.