First page Back Continue Last page Overview Graphics
Once there was IPv4
Network Address Translation (NAT)
- Main workaround for global address depletion
- Breaks end-to-end model
- NAT aware applications or application aware NAT
- Need for third party registration servers to find peers
Other problems with IPv4
- Not tamper-proof, no encryption at network level, no mobility
Notes:
Let’s start from the beginning. Once there was IPv4, a protocol designed over 30 years ago by the DoD. It is the predominant network layer protocol in the Internet today. The biggest problem we face with IPv4, is the lack of globally unique addresses to assign to devices. IPv4, does not provide enough addresses to the meet the growing needs of the global network.
Ex:
- an Internet late country requests address space, gets one class C network. 254 computers for the entire country?
- a medium size organization with 5000 thousand employees requests address spaces, gets 32 addresses.
The US has 70% of the entire address space, that’s why we don’t really have a big problem (we are feeling the lack of addresses, but nowhere near as much as Asian or European countries). To cope with the address depletion, two mechanism were developed: NAT, and CIDR.
CIDR allowed for more flexible and less wasteful allocation of IP addresses. NAT is a mechanism that allows the creation of “private” networks, and translates addresses between private and public addresses.
NAT has a number of problems: the communication is not end-to-end, there is now a single point of failure that must be configured and maintained. Applications need to be rewritten to make them NAT-aware, or NATs need to be upgraded each time a new application is deployed to make them application aware. There is no way for clients to find each other without third-party servers, this blocks the deployment of P2P applications.
Some other problems with IPv4, arise from the times it was designed in. There protocol has no consideration for mobility or integrated security.