Web Site

Computerit-solutions.com



» Computer » Computer network » Topics begins with I » IP fragmenting


Page modified: Friday, June 23, 2006 20:28:28

The IP fragmenting designates the allocation of a IP packet on several physical data blocks, if the overall length of the packet is larger than the maximum transfer unit of the network interface.

Background

In the simplest case the entire IP datagram fits into a physical data block and achieves thus the highest efficiency. There are however no firm defaults concerning a firm package size of IP packets. The moreover one the maximally possible package size depends on the used infrastructure components, since each package Switching technology exhibits different maximum package sizes.

Objective

A goal during the introduction of fragmenting in IP minutes was it to hide (layer model) and thus the minutes implementation arrange independent of the hardware the underlying network architecture for the user.

Function

As soon as the IP stack (see also OSI model or TCP/IP reference model) contains a packet for dispatching, this examines whether the package size makes an allocation necessary on the basis the MTU given for the network interface which can be used. If this is necessary, then this divides the existing packet into several packets. This procedure is called fragmenting. This fragmenting can take place both at the original transmitter or also on routing, which lies between transmitters and receivers. If a IP datagram is fragmented, then it is only built up with the receiver again (exception: Firewalls if necessary zwischengschaltete was particularly instructed to lead, a so-called Reassembly through, before the data are continued to lead). If it should be necessary, also a package already fragmented can be further fragmented (for instance with a change of the transmission technique).

Each datagram, which was fragmented, receives a marking (to fragment flag, see IPv4) and contains instead of the datagram header the original package a so-called fragment header, which indicates among other things the so-called offset of the data portion dispatched in this package in relation to the original package.

By definition the IP layer cannot give data concerning whether a package in the process of its transmission is fragmented or not. Only exception: The transmitter can set that for so-called Don"´t fragment flag, which instructs all communication systems involved (rout, gateway, etc.) to make no fragmenting. If a fragmenting were nevertheless necessary, the package is rejected and sent to the transmitter ICMP an error message that the goal is not attainable - "“fragmentation required but don't fragment bit set"”.

Since each individual fragment represents its own package, it can also occur that these individual parts do not arrive arranged. Here it is then the task of the receiver to set the original from the information existing in the Paketheadern again together.

Effects

Although the objective one for higher layers (e.g. TCP/UDP) transparent implementation is, gives it two points, in which this is reached not completely:

  • Fragmenting can have large influence on the performance and the information flow-rate and affects this generally negatively.
  • If a fragmented package of the original package is lost, then the entire original must again be transferred. IP has however no safeguard and/or Timeoutmechanismen and is here dependent on the security functions of higher layers like those of the TCP.

For reasons mentioned is tried to avoid fragmenting always so far as possible.


Related Websites

We found here 6 related websites.

Page cached: Wednesday, July 5, 2006 14:10:27
Valid XHTML 1.0!  Valid CSS!

Page copy protected against web site content infringement by Copyscape