cancel
Showing results for 
Search instead for 
Did you mean: 
cancel
608
Views
10
Helpful
5
Replies

Why should we use udp?

huangzhy
Level 1
Level 1

Netware has a transport layer service similar to tcp called SPX,but no service similar to UDP.Applications requiring connectionless service directly access the connectionless IPX at the network layer.What purpose does UDP serve by providing a connectionless service on top of what is already a connectionless service?

5 Replies 5

According to TCP/IP Model, Tcp and Udp protocols are located at transport layer(not OSI).Tcp requires connection service because ,while sending packets,packets are numbered,and arrive sequentially.Furthermore security is important for tcp(while using ftp,http,smtp and dns).On the other hand, udp doesnot requires connection service thanks to this,security is not important and packets are not numbered.(For example television broads in the internet)Udp usually is used for DNS and TFTP

What you explained had no relationship with what i asked ,i think.

If you are asking what is the point of UDP I think I may be able to help. UDP provides two services not provided by the IP layer. It provides port numbers to help distinguish different user requests and, optionally, a checksum capability to verify that the data arrived intact. UDP is a good alternative to TCP as it is connectionless, therefore doesn't require the overhead and added traffic load of acknowledgements. TCP packets have a header of 20 bytes (without options) while UDP has an 8 byte header. If traffic isn't seen as vital enough (ie best effort is good enough) to require an ack, why have the added load/traffic of acks. Because UDP has eliminated the need for acks, it is fast and efficient. TCP connections are slower to establish and require more overhead. UDP packets can be sent without requiring the device to wait for an ack or keep the memory needed for it. UDP therefore has a valuable role.

Hope it helps.

Steve

Yes, so many time-sensitive application will use UDP instead of TCP, such as RTP(Real Time protocols).

vcjones
Level 5
Level 5

UDP provides two critical services: multiplexing (through the port numbers) and error detection (for corrupted data using a checksum). The former allows multiple processes to independently communicate between the same two machines. The latter detects reassembly failures and other gross errors not caught by the error checking on the link level. Both of these functions are built into the version of the IPX protocol used by Novell. Novell IPX has a 16 bit checksum at the network layer and IPX addresses consist of a 32 bit network number (used for routing between networks), a 48 bit machine number (to distinguish multiple machines on the same network), and a 16 bit local address (the equivalent of a UDP socket).

UDP also includes a data length indicator which allows the receiver to accurately determine the end of the UDP payload. This is normally redundant information, but could be used to pack multiple UDP packets into a single IP packet or to use UDP over a streaming network protocol. IPX also includes a length field to serve the same purpose.

Consequently, in the Novell implementation of IPX, there is no need for a UDP equivalent protocol because the UDP services are already provided by the IPX protocol.

A historical note--traditionally the IPX checksum has been disabled "to improve performance." This attitude changed dramatically about 10 years ago when a defect in a popular Ethernet card allowed the delivery of corrupted frames. When faced with the reality of needing to choose between a minor improvement in speed and preserving the integrity of data in the file system, there was a mad rush to turn IPX checksums back on!

Good luck and have fun!

Vincent C Jones

www.networkingunlimited.com

Getting Started

Find answers to your questions by entering keywords or phrases in the Search bar above. New here? Use these resources to familiarize yourself with the community: