What are IP fragments?
The IP protocol specification provides guidance for performing IP fragmentation – a process by which IP datagrams are subdivided into smaller data packets during transit. Fragmentation is needed because every network architecture carries data in chunks called frames, and the maximum frame size varies from network to network. When an IP datagram enters a network whose maximum frame size is smaller than the size of the datagram, it is split into fragments. Thereafter, the fragments travel separately to their destination, at which point they are reassembled and processed. This vulnerability results because of a flaw in the way the affected systems perform IP fragment reassembly. If a stream of IP fragments with a particular type of malformation are directed against an affected machine, the work factor associated with performing IP fragment reassembly can be driven arbitrarily high by varying the data rate at which the fragments are sent. This could allow a malicious user to consume most o
