分类：
2012-12-20 11:11:11
Link Layer is most basic. Its function is ensure that if a device sends a packet of information CallManager, or CallManager sends a packet of information to a device, sent packet is received. CallManager uses two methods of communication. Transmission Control Protocol (TCP) is by far the most commonly used. TCP underlies much communication on the Internet. It provides for reliable communication between peers using the Internet Protocol. CallManager uses TCP for call signaling and media control with CallManager nodes, media devices, IP phones, H.323 gateways, and ISDN call signaling originating from MGCP gateways. The User Datagram Protocol is a protocol in which a sent packet is not guaranteed to be received. CallManager uses UDP for communication with MGCP gateways and SIP proxies. Although UDP itself is not reliable, MGCP or SIP is designed to handle instances where the IP network loses the message; in such a case, MGCP or SIP retransmit its last message.
The Protocol Layer includes the logic that CallManager uses to manage the different types of devices that it supports. These devices include media devices, trunk devices, and station devices. The Protocol Layer also supports third-party integration with CallManager through the TAPI and JTAPI protocols.
The Aggregator Layer allows CallManager to properly handle the interactions between groups of related devices. The media resource manager, for example, permits one CallManager node to locate available media devices, even if they are registered to other CallManager nodes. The route list performs a similar function for gateways. Line control permits CallManager to handle IP phones that share a line appearance, even if the IP phones are registered with different CallManager nodes.
The Media Control Layer handles the actual media connections between devices. It handles the media control portion of setting up a call, but it also handles more complicated tasks. For instance, sometimes CallManager must introduce a transcoding device to serve as an interpreter for two devices that don't communicate via the same codec. In this case, one call between two devices consists of multiple media hops through the network. The Media Control Layer coordinates all the media connections.
The Call Control layer handles the basic call processing of the system. It locates the destination that a caller dials and coordinates the Media Control, Aggregator, and Protocol Layers. Furthermore, it provides the primitives that the Supplementary Service Layer uses to relate independent calls. The Supplementary Service Layer relates independent calls together as part of user-requested features such as transfer, conference, and call forwarding.
Within each layer, the SDL application engine manages state machines, which are essentially small event-driven processes, but they do not show up on the Microsoft Windows 2000 Task Manager. Rather, the SDL application engine manages state machine tasks. These state machines each handle a small bit of the responsibility of placing calls in a CallManager network. For example, one kind of state machine is responsible for handling station devices, whereas another type is responsible for handling individual calls on station devices.
These state machines perform work through the exchange of proprietary messages. Before CallManager release 3.0 was created, these messages were strictly internal to CallManager. With the 3.0 release, these messages could travel from a state machine in one CallManager node directly to another state machine managed by a different CallManager node. This mechanism is, in fact, what allows a CallManager cluster to operate with perfect feature transparency. The same signaling that occurs when a call is placed between two devices managed by the same CallManager node occurs when a call is placed between two devices managed by different CallManager nodes.
Architecturally, intracluster communication tends to occur at the architectural boundaries listed in. Take, for example, the situation that occurs when two devices that share a line appearance register with different CallManager nodes. When someone dials the directory number of the line appearance, both devices ring. Even though the state machine responsible for managing each station is on its own CallManager node, both of these state machines are associated with a single state machine that is responsible for managing line appearances. (These can reside on one of the two CallManager nodes in question, or possibly on a third CallManager node.) The ICCS, however, guarantees that the feature operates the same, no matter how many CallManager nodes are handling a call.
The architectural layers are rather loosely coupled. In theory, a call between two devices registered to different CallManager nodes in the cluster could involve up to seven CallManager nodes, although in practice, only two are required.
