tcp-ip IP( Internet Protocol )

IP provides an unreliable, connectionless datagram delivery service.

By unreliable we mean there are no guarantees that an IP datagram successfully gets to its destination. When something goes wrong IP has a simple error handling algorithm: throw away the datagram and try to send an ICMP message back to the source. Any required reliability must be provided by the upper layers (e.g., TCP).

The term connectionless means that IP does not maintain any state information about successive datagrams. Each datagram is handled independently from all other datagrams. This also means that IP datagrams can get delivered out of order. If a source sends two consecutive datagrams (first A, then B) to the same destination, each is routed independently and can take different routes, with B arriving before A.

IP Record Route Option

Every router that handles the datagram to add its IP address to a list in the options field. The biggest problem, however, is the limited room in the IP header for the list of IP addresses.

The header length in the IP header is a 4-bit field, limiting the entire IP header to 15 32-bit words (60 bytes). Since the fixed size of the IP header is 20 bytes, and the RR option uses 3 bytes for overhead, this leaves 37 bytes (60-20-3) for the list, allowing up to nine IP addresses. 

Type is a I-byte field specifying the type of IP option. For the RR option its value is 7. Len is the total number of bytes of the RR option and it's maximum size is 39, to record up to nine IP addresses.

Ptr is called the pointer field. It is a 1-based index into the 39-byte option of where to store the next IP address. Its minimum value is 4, which is the pointer to the first IP address. As each IP address is recorded into the list, the value of ptr becomes 8, 12, 16, up to 36. After the ninth address is recorded ptr becomes 40, indicating the list is full.

When a router (which by definition is multihomed) records its IP address in the list, the router records the outgoing IP address.

IP Timestamp Option

The type field is 0x44 for the timestamp option. The two fields len and ptr are the same as for the record route option: the total length of the option (normally 36 or 40) and a pointer to the next available entry (5,9, 13, etc.).

The next two fields are 4-bit values: OF is the overflow field and FL is a flags field. The operation of the timestamp option is driven by the flags field

flagsDescription0Record only timestamps1Each router records its IP address and its timestamp. There is room for only four of these pairs in the options list.3The sender initializes the options list with up to four pairs of IP addresses and a 0 timestamp. A router records its timestamp only if the next IP address in the list matches the router's.

If a router can't add a timestamp because there's no room left, it just increments the overflow field.

IP Source Routing Option

The idea behind source routing is that the sender specifies the route. Two forms are provided:

• Strict source routing. The sender specifies the exact path that the IP datagram must follow. If a router encounters a next hop in the source route that isn't on a directly connected network, an ICMP "source route failed" error is returned.
• Loose source routing. The sender specifies a list of IP address that the datagram must traverse, but the datagram can also pass through other routers between any two addresses in the list.

The code is 0x83 for loose source routing, and 0x89 for strict source routing.

The source route options are actually called "source and record route" (LSRR and SSRR, for loose and strict) since the list of IP addresses is updated as the datagram passes along the path. What happens is as follows:

1. The sending host takes the source route list from the application, removes the first entry (it becomes the destination address of the datagram), moves all the remaining entries left by one entry, and places the original destination address as the final entry in the list. The pointer still points to the first entry in the list (e.g., the value of the pointer is 4).
2. Each router that handles the datagram checks whether it is the destination address of the datagram. If not, the datagram is forwarded as normal. (In this case loose source routing must have been specified, or we wouldn't have received the datagram.)
3. If the router is the destination, and the pointer is not greater than the length, then (1) the next address in the list (where ptr points) becomes the destination address of the datagram, (2) the IP address corresponding to the outgoing interface replaces the source address just used, and (3) the pointer is incremented by 4.

IP Fragmentation

When an IP datagram is fragmented, it is not reassembled until it reaches its final destination. The IP layer at the destination performs the reassembly. The goal is to make fragmentation and reassembly transparent to the transport layer (TCP and UDP), which it is, except for possible performance degradation. It is also possible for the fragment of a datagram to again be fragmented (possibly more than once). The information maintained in the IP header for fragmentation and reassembly provides enough information to do this.

Recalling the IP header, the following fields are used in fragmentation. The identification field contains a unique value for each IP datagram that the sender transmits. This number is copied into each fragment of a particular datagram. (We now see the use for this field.) The flags field uses one bit as the "more fragments" bit. This bit is turned on for each fragment comprising a datagram except the final fragment. The fragment offset field contains the offset of this fragment from the beginning of the original datagram. Also, when a datagram is fragmented the total length field of each fragment is changed to be the size of that fragment.

Finally, one of the bits in the flags field is called the "don't fragment" bit. If this is turned on, IP will not fragment the datagram. Instead the datagram is thrown away and an ICMP error ("fragmentation needed but don't fragment bit set") is sent to the originator.

If one fragment is lost the entire datagram must be retransmitted.

Fragmentation requires that the data portion of the generated fragments (that is, everything excluding the IP header) be a multiple of 8 bytes for all fragments other than the final one.