SIP
Session Initiation Protocol (SIP) is the IETF protocol for IP telephony. It only supports IP-based phones. It has a smaller footprint than H.323 so it's faster and more scalable. The problem lies in the fact that it's a newer protocol, and therefore fewer products exist that use it. However, SIP addresses some of the shortcomings of H.323 by making users easier to identify, making it easier to connect two circuit-switched networks across an IP network, and decreasing the delay in call setup time.
SIP identifies users with a Hierarchical URL. This URL is based on a user's phone number or host name and looks similar to an e-mail address (for example,SIP: [email protected]). Figure below illustrates the SIP call process.

9
When a call is made, the caller initiates it with an invite request. This request contains the information necessary for the person you're calling to join the session: the media types and formats for the call, the destination for the media data, and perhaps requests for using H.261 video and/or G.711 audio. The invite request is sent to the user's SIP server. Because you include your available features in the invite request, the negotiation of the connection takes place in a single transaction thus call setup time is decreased (approximately
100 milliseconds).
The SIP server may or may not be a proxy server. A SIP proxy server
receives the request and figures out the user's location using its internal algorithms.
A non-proxy SIP server functions as a redirect server in that it sends
back to the user the SIP URL that the user uses to query. In both the redirect and proxy server cases, the server's address is obtained by querying the Domain Name Service (DNS).
Once the SIP URL is found, the request finally makes it to the person
you're trying to call. If the person picks up the call, the receiver's client responds to the invite request with the capabilities of its software (videoconferencing, whiteboarding,and so forth), and the connection is established.
SIP has two features that really make it unique:
_
It can split an incoming call so that multiple extensions can be rung at once. When the invite request comes in, the SIP server can return to the initiator of the call a Web Interactive Voice Response (IVR) page, which contains extensions of different departments or users in a list.
All you have to do is click on the link to call the appropriate person or department.
_
It can return different media types.
SIP is simple and easy to deploy because its only job is to identify the user and set up the call; it relies on other protocols and applications to manage the call. It utilizes existing DNS instead of having to create a separate database for telephony. It also interfaces with circuit-switched networks (the PSTN) more easily than H.323. Does this mea SIP is the way to go? Not necessarily.
It is not widely available, and (the biggest drawback at this point) it must
"de-throne" Microsoft. Every version of Windows that ships has an H.323 client as part of the package (it's free!). Whether a company will purchase another client all depends on its needs and what it wants to accomplish with IP telephony.
While H.323 is the accepted VoIP protocol today, many people think that SIP will be the VoIP protocol of the future. Most of the larger vendors are developing SIP-based solutions if they haven't already. It will be beneficial to understand both protocols to make a decision on what kind of VoIP solution to deploy.