IPv6 routers don’t split oversized packets as they travel. Path MTU Discovery helps the sender find the route’s size limit, while an ICMPv6 Packet Too Big message reports a smaller MTU when needed. The sender can then fragment the packet or send smaller ones.

Multiple Choice

How is fragmentation handled in IPv6, and what happens if a packet is too large?

In IPv6, fragmentation is handled by the sender, not by routers along the path. The network relies on the Path MTU Discovery process to learn the smallest MTU along the route, and the source must ensure packets fit that size before sending. If a router cannot forward a packet because it’s larger than the next hop’s MTU, it will drop the packet and return an ICMPv6 “Packet Too Big” message to the sender, indicating the MTU to use. The sender can then fragment (at the source) or retransmit smaller packets that conform to the discovered path MTU. The fragmentation header in IPv6 exists but is not used by routers to fragment in transit; fragmentation is essentially a source-driven operation, with PMTUD guiding the allowed size.

How IPv6 Handles Packets That Are Too Large

IPv6 takes a different approach to fragmentation than IPv4. When an IPv6 packet is too large for a link along its route, a router doesn’t split it into smaller pieces. Instead, it drops the packet and sends a message back to its source. The sender can then adjust the packet size or divide the data before sending it again.

That may sound like a small change, but it shifts an important job from routers to the devices communicating across the network. It also shapes how IPv6 discovers the right packet size for a route, and what happens when that discovery goes wrong.

Routers forward; they don’t fragment

Every network link has a maximum transmission unit, or MTU: the largest packet that link can carry in one piece. A route across the internet may pass through several links, each with its own MTU. The smallest MTU on the path is called the path MTU.

With IPv4, a router could fragment a packet that was too large for the next link, unless the packet’s “don’t fragment” setting prevented it. IPv6 removes that option for routers. They forward packets as received, but if a packet exceeds the next link’s MTU, they discard it.

Why put the burden on the sender? Routers handle huge volumes of traffic. Asking them to inspect and split packets takes time and resources, and creates more work at every hop. IPv6 instead lets the source learn about the path and send packets sized to fit. The middle of the network stays simpler.

This doesn’t mean IPv6 has no fragmentation. It means fragmentation is a source-driven operation. If needed, the sending device can split the original data into fragments and include an IPv6 Fragment extension header. Routers still won’t split those fragments further.

The message that tells the sender what went wrong

When a router can’t forward an oversized IPv6 packet, it sends an ICMPv6 Packet Too Big message to the source. This is ICMPv6 Type 2. The message includes the MTU of the link that couldn’t carry the packet, giving the sender useful information rather than just reporting a failure.

The sender can use that information to lower the size of its packets. It might send smaller packets from then on, or divide a larger piece of data into fragments at the source. In many cases, the transport protocol or application adjusts how much data it sends at once, so IPv6 fragmentation may not be needed.

A key point: the router drops the oversized packet before sending the notice. The Packet Too Big message isn’t a request for the router to retry with smaller pieces. It’s a signal to the source: “This packet won’t fit here; send something smaller.”

Path MTU Discovery, step by step

IPv6 uses Path MTU Discovery (PMTUD) to help a sender find the largest packet size that can travel across a route without being dropped. The basic process works like this:

  1. The sender transmits packets at a size it believes the path can handle.

  2. If a router finds a packet too large for the next link, it drops it and sends an ICMPv6 Packet Too Big message.

  3. The sender reduces its packet size and sends again.

  4. Over time, the sender settles on a size that fits the path.

It’s a bit like moving a sofa through a building: if it won’t fit through one doorway, you need to change your plan before trying again. With packets, the router’s message provides the measurement—the MTU to respect.

The basic IPv6 link MTU must support at least 1280 bytes. That minimum helps ensure devices can communicate across IPv6 links, though some links can carry larger packets. A sender still needs to account for the actual path, not just the MTU of its own local connection.

Fragmenting at the source

If a sender needs to split a packet, it adds a Fragment header to the IPv6 header chain. The header carries information used to reassemble the pieces, including a fragment offset and an identification value. The final fragment is marked so the receiver can tell that no more pieces are coming.

The destination—not a router along the way—reassembles the fragments. If one fragment goes missing, the full original packet can’t be reassembled. That can lead to delays or retransmission, depending on the transport protocol and the application.

Fragmentation also adds overhead. Each fragment carries headers, and the receiver must hold and reassemble the pieces. That’s why sending data in appropriately sized packets is often preferable to splitting oversized packets. For many applications, transport-layer segmentation—such as TCP sending data in manageable chunks—handles the job without using the IPv6 Fragment header.

The Fragment header is part of IPv6’s extension-header system. It doesn’t sit at a fixed “end” of every IPv6 packet; it appears in the header chain when the source chooses to fragment. Ordinary IPv6 packets don’t need one.

When Packet Too Big messages go missing

PMTUD depends on the sender receiving the ICMPv6 Packet Too Big message. If a firewall or network device blocks that message, the sender may keep transmitting packets that are too large. The packets disappear, but the sender never gets a clear explanation. This can create a PMTU black hole: a connection may partly work, then stall when it tries to send larger packets.

That’s one reason ICMPv6 matters. It isn’t merely optional diagnostic traffic; IPv6 relies on several ICMPv6 messages for normal network functions. Blocking all ICMPv6 can break things in ways that are hard to spot. Network filtering should allow the messages needed for proper operation while still applying sensible security controls.

There’s also a related technique called Packetization Layer Path MTU Discovery, or PLPMTUD. Rather than relying only on network-layer ICMP messages, it lets a transport protocol or application infer whether packets are getting through, often using acknowledgments and carefully sized probes. It can help when ICMP messages are lost or filtered. Still, it complements the broader goal: find a packet size the path can carry.

Why this design matters

Keeping fragmentation out of routers makes IPv6 forwarding more predictable. Each router can focus on sending packets onward, rather than breaking them apart and adding processing overhead. The source has more information about the communication and can decide how to package its data.

There is a trade-off, of course. The sender must respond when the path MTU is smaller than expected, and the feedback system has to work. But that responsibility sits with the device best placed to adapt its traffic. Once the path size is known, subsequent packets can travel more efficiently.

So the central rule is simple: IPv6 routers do not fragment packets in transit. If a packet is too large, the router drops it and returns an ICMPv6 Packet Too Big message, including the relevant MTU. The sender can then reduce packet sizes or fragment the data at its own end. That design keeps the network core lean—and gives the source the information it needs to get the packet through.