CODEWITHSUNDEEP
socketsnetwork-programmingtcp
jdogdvr
jdogdvr

Reputation: 369

How websockets work in respect to TCP/IP and HTTP?

Hi guys I'm new to understanding protocols used over the web and need some help in understanding the basics of websockets,TCP/IP and HTTP.

My understanding of the relation between TCP/IP and HTTP is that IP is required to connect all networks. TCP is a mechanism that allows us to transfer data safely and HTTP, which utilizes TCP to transfer its data, is a specific protocol used by Web servers and clients.

Does this mean you can't send a HTTP request without TCP?

Websockets communicate using TCP layer and a connection between client and server is established through HTTP which is known as the handshake process.

Does websockets have its own protocol? How can you send a http request(hand shake process) to establish TCP/IP when you need TCP to carry out an HTTP request. I know I am missing something really important here and would be great to get my understanding of these protocols sharpened!

Upvotes: 1

Views: 1600

Answers (1)

Winner1235813213455
Winner1235813213455

Reputation: 360

Firstly, IP is not necessarily required to connect all networks. However, it is the most widely used and adopted today (For now that is). Old network protocols such as Appletalk, IPX, and DECNet are all legacy network protocols that are not much used anymore, however they are still around to an extent. Don't forget IPv6 is out there as well, in some places, and can ride over IPv4 networks if your configuration is done properly.

When you say TCP being "safe", I would give it another word, and that would be intelligent. TCP is a transport protocol, and is the header that comes directly after the IPv4 header. TCP is primarily used for flow control and has become very efficient at error recovery in case a part of a packet or packets has been last when transferring/receiving. While this is great for some transactions, the error control requires an additional amount of overhead in the packet. Some applications, let's say VoIP for example, is very sensitive to delay, jitter (Variation in delay) and congestion. This is why it uses UDP.

Like TCP, UDP is a transport protocol, however there is no flow control. Think of it this way: When sending packets over TCP, it's like asking the other end if they received your message. If they did, they will acknowledge it. If not, you now have to determine how you will resend this information. UDP has none of this. You send your message to the other side, and hope it gets there.

Now if you want to talk about "safe" protocols, this is usually done at either the network layer (IPSec) or the application layer (SSL). Safe typically means secured.

A usual TCP three-way handshake looks like this:

Whoever sends the SYN is the client. Whoever receives that initial SYN is the server.

Client sends SYN --> Server

Now, if the server is listening, and/or there's not a firewall blocking the service (Which in that case you'd receive a TCP frame from the server with the RST,ACK bits set most likely), the server will respond with a SYN-ACK:

Server sends SYN/ACK --> Client

If the client received this packet, he will acknowledge he received it. This completes the three-way handshake and these two may begin exchanging information.

Client sends ACK --> Server

Here's a good site for some info:

http://www.tcpipguide.com/free/index.htm

Upvotes: 2

Related Questions