1. From two-level IP addresses to three-level IP addresses
In today's opinion, in the early days of ARPANET, the design of IP addresses was indeed unreasonable.
First, the utilization of IP address space is sometimes very low.
The number of hosts that can be connected to each class A address network exceeds 10 million, and the number of hosts that can be connected to each class B address network is also more than 60,000. Some units have applied for a class B address network, but the number of connected hosts is not large, but they are unwilling to apply for a class C address that is sufficient for use, because the reason is that it may be developed in the future. The waste of IP addresses will also cause the resources of the IP address space to be used up prematurely.
Second, assigning a network number to each physical network would make the routing table too large and thus degrade network performance.
Every router should be able to find out from the routing table how to get to the next hop router on other networks. Therefore, the more networks there are in the Internet, the more entries the router's routing table has. In this way, even if we had enough IP address resources to assign a network number to each physical network, it would result in too many entries in the router's routing table. This not only increases the cost of the router (requires more storage space), but also takes more time to find the router, and also dramatically increases the routing information that is regularly exchanged between routers, thus reducing the performance of the router and the entire Internet. going down.
Third, two-level IP addresses are not flexible enough.
Sometimes in emergencies, a unit needs to immediately start a new network at a new location. But before applying for a new IP address, it is impossible for the newly added network to connect to the Internet to work. We hope to have a method that enables a unit to flexibly increase its network at any time without having to go to the Internet management agency to apply for a new network number in advance. The original two-level IP address couldn't do this yet.
In order to solve the above problems, a "subnet number field" has been added to the IP address since 1985 , so that the two-level IP address becomes a three-level IP address. Very flexible. This method is called subnetting, or subnet addressing or subnet routing . Subnetting has become an official standard protocol for the Internet.
The basic idea of subnetting is as follows:
(1) A unit with many physical networks can divide the physical network it belongs to into several subnets . Subnetting is purely an intra-organizational matter. The network outside the unit cannot see how many subnets the network is composed of, because the unit still appears as a network to the outside world.
(2) The method of dividing the subnet is to borrow several bits from the host number of the network as the subnet number. Of course, the host number is reduced by the corresponding number of digits. Therefore, the two-level IP address becomes a three -level IP address within the unit : the network number, the subnet number, and the host number. It can also be expressed in the following notation:
IP address::={<network number>,<subnet number>,<host number>}
(3) All IP datagrams sent from other networks to a host in the unit are still based on the destination network number of the IP datagram to find the routers that are continuously on the unit's network. However, after receiving the IP datagram, the router finds the destination subnet according to the destination network number and subnet number, and delivers the IP datagram to the destination host.