H.323 Network Components

H.323 has several functional components, which you can implement separately on different devices, or as multiple functions on the same device. The basic components include gateways, gatekeepers, terminals, and multipoint control units (MCU). You can also use proxy servers. Figure 3-1 illustrates these components. A Cisco CallManager is also shown. It is not part of the H.323 specification, but it can interact with H.323 devices.

Figure 3-1. H.323 Network Components

H.323 Gateways

H.323 gateways translate between different types of networks, just as any gateway does. In Figure 3-1, the gateways use H.323 to communicate with each other, the gatekeeper, CallManager, and the other H.323 devices in the network. H.323 gateways can register with a gatekeeper. They can also register analog phones in their network with the gatekeeper. H.323 gateways have the intelligence to place and receive calls, although CallManager can also control calls. Dial peers are configured on the gateway, with destination patterns. A Voice over IP (VoIP) dial peer points to the CallManager for each range of internal phone numbers it controls. CallManager sends outgoing PSTN calls to the gateway, which uses plain old telephone service (POTS) dial peers with the appropriate destination patterns to route them.

H.323 Gatekeepers

Gatekeepers provide a centralized point to resolve E.164 phone numbers to IP addresses and do call admission control. They can provide call routing and control, security, and bandwidth management, and they can simplify dial peer configuration. In Figure 3-1, when a call needs to go to an IP phone in the other network, the gateway must contact the gatekeeper before the call can proceed.

When people make calls between IP phones that the same CallManager cluster controls, the CallManager maps the called phone number to a destination IP address. When they make calls between clusters, they can optionally use an H.323 gatekeeper to route the call to the appropriate gateway or CallManager. Both CallManager and H.323 gateways can use the services of a gatekeeper. The originating gateway (or CallManager) sends a request to the gatekeeper to admit the call, and the gatekeeper responds with the IP address of the terminating gateway or CallManager. Call setup and voice media can then be sent directly between the originating and terminating peers.

H.323 Terminals

Like MGCP, H.323 uses the concept of endpoints. IP phones and video conferencing stations, in addition to gateways and CallManagers, can be H.323 endpoints. An H.323 terminal is an endpoint that is able to do real-time two-way communication. All H.323 terminals must support H.225 for call setup, H.245 for channel and capability control, RAS for communicating with a gatekeeper, and Real-Time Transport Protocol (RTP)/Real-Time Transport Control Protocol (RTCP) for media streams. They can optionally support the other H.323 specifications. The most common type of H.323 terminal is a video conferencing system.

Multipoint Control Units

An MCU allows multiple participants in a conference. In Figure 3-1, if the three H.323 video systems want to participate in a conference, the MCU controls that. An MCU is composed of a multipoint controller (MC) and a multipoint processor (MP). The MC handles the H.245 capabilities negotiation and controls the conference resources. The MP does the actual mixing and splitting of the audio and video streams, and translation between different codecs or bandwidth.

H.323 Proxy Servers

H.323 proxy servers do the call setup and teardown in place of the endpoint. They can provide security, quality of service (QoS), and application-specific routing (ASR). In Figure 3-1, a Cisco router is configured as an H.323 proxy server. It can communicate with the gatekeeper and endpoints in the other network, thus hiding their real IP address of endpoints in its network. Two proxy servers can implement Resource Reservation Protocol (RSVP) for calls between them, in proxy for their non-RSVP aware endpoints. With ASR, a proxy server routes traffic based on the application used, rather than the IP address. For instance, voice traffic could be directed to a special QoS-enabled path, instead of taking the same path as data.