number portability, on the Internet such a service is not required in the
first place, since URIs have no geographic significance. Caller ID is
another paid “service” that makes no sense for SIP, since just like in
e-mail, the To:and From:headers are always there without extra cost.
■■ SIP is not a transfer protocol such as HTTP, designed to carry large
amounts of data. It is designed to transport only small amounts of data
required to set up interactive communications. Small amounts of data
not related to call setup (such as short text messages for instant mes-
sages) are well suited for SIP, as will be shown in Chapter 13, “Presence
and Instant Messaging,” but large amounts of general data are not
suited for carrying by SIP.
■■ SIP is not a resource reservation or prioritization protocol, so it cannot
ensure QoS but can only interwork with other protocols designed to
support QoS, as will be discussed in Chapter 18, “Quality of Service for
Real-Time Internet Communications.”Divergent Views on the Network
No book on SIP would be complete without mentioning the fact there are com-
pletely divergent views on the network and how SIP will be used [34].
The Internet view on the network is as follows:
■■ The Internet is thenetwork. The next generation network (NGN) is
IPv6.
■■ The Internet is transparent e2e or just “dumb;” it is application-unaware.
■■ User consent and control resides in the endpoints.
■■ Service availability is what matters to users and not QoS. QoS is good
as long as network congestion is avoided and, if so, voice quality is an
endpoint capability.
■■ The Internet is the result of a continuous evolution, and the architecture
changes constantly over time [35].
The ITU-T view on the network is as follows:
■■ The NGN will be derived form the PSTN, but using IP technology; the
IP Multimedia System (IMS).
■■ The NGN is application-aware.
■■ Control resides in the network.
■■ The NGN has ample QoS definitions and guarantees for the network
service.34 Chapter 2