1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
|
<?xml version="1.0" encoding="utf-8" ?>
<!DOCTYPE chapter SYSTEM "chapter.dtd">
<chapter>
<header>
<copyright>
<year>2000</year><year>2013</year>
<holder>Ericsson AB. All Rights Reserved.</holder>
</copyright>
<legalnotice>
Licensed under the Apache License, Version 2.0 (the "License");
you may not use this file except in compliance with the License.
You may obtain a copy of the License at
http://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing, software
distributed under the License is distributed on an "AS IS" BASIS,
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
See the License for the specific language governing permissions and
limitations under the License.
</legalnotice>
<title>Architecture</title>
<prepared>Håkan Mattsson</prepared>
<responsible>Håkan Mattsson</responsible>
<docno></docno>
<approved>Håkan Mattsson</approved>
<checked></checked>
<date>2007-06-15</date>
<rev>%VSN%</rev>
<file>megaco_architecture.xml</file>
</header>
<section>
<title>Network view</title>
<p>Megaco is a (master/slave) protocol for control of gateway functions at
the edge of the packet network. Examples of this is IP-PSTN trunking
gateways and analog line gateways. The main function of Megaco is to
allow gateway decomposition into a call agent (call control) part (known
as Media Gateway Controller, MGC) - master, and an gateway interface
part (known as Media Gateway, MG) - slave. The MG has no call control
knowledge and only handle making the connections and simple
configurations.</p>
<p>SIP and H.323 are peer-to-peer protocols for call control (valid only
for some of the protocols within H.323), or more generally multi-media
session protocols. They both operate at a different level (call control)
from Megaco in a decomposed network, and are therefor not aware of
whether or not Megaco is being used underneath.</p>
<image file="megaco_sys_arch.gif">
<icaption>Network architecture</icaption>
</image>
<p>Megaco and peer protocols are complementary in nature and entirely
compatible within the same system. At a system level, Megaco allows
for</p>
<list type="bulleted">
<item>
<p>overall network cost and performance optimization</p>
</item>
<item>
<p>protection of investment by isolation of changes at the call
control layer</p>
</item>
<item>
<p>freedom to geographically distribute both call function and
gateway function</p>
</item>
<item>
<p>adaption of legacy equipment</p>
</item>
</list>
</section>
<section>
<title>General</title>
<p>This Erlang/OTP application supplies a framework for building
applications that needs to utilize the Megaco/H.248 protocol.</p>
<p>We have introduced the term "user" as a generic term for either
an MG or an MGC, since most of the functionality we support, is
common for both MG's and MGC's. A (local) user may be configured
in various ways and it may establish any number of connections
to its counterpart, the remote user. Once a connection has been
established, the connection is supervised and it may be used for
the purpose of sending messages. N.B. according to the standard
an MG is connected to at most one MGC, while an MGC may be
connected to any number of MG's.</p>
<p>For the purpose of managing "virtual MG's", one Erlang node may
host any number of MG's. In fact it may host a mix of MG's and
MGC's. You may say that an Erlang node may host any number of
"users".</p>
<p>The protocol engine uses callback modules to handle various
things:</p>
<list type="bulleted">
<item>
<p>encoding callback modules - handles the encoding and
decoding of messages. Several modules for handling different
encodings are included, such as ASN.1 BER, pretty well
indented text, compact text and some others. Others may be
written by you.</p>
</item>
<item>
<p>transport callback modules - handles sending and receiving
of messages. Transport modules for TCP/IP and UDP/IP are
included and others may be written by you.</p>
</item>
<item>
<p>user callback modules - the actual implementation of an MG
or MGC. Most of the functions are intended for handling of a
decoded transaction (request, reply, acknowledgement), but
there are others that handles connect, disconnect and
errors cases.</p>
</item>
</list>
<p>Each connection may have its own configuration of callback
modules, re-send timers, transaction id ranges etc. and they may
be re-configured on-the-fly.</p>
<p>In the API of Megaco, a user may explicitly send action
requests, but generation of transaction identifiers, the
encoding and actual transport of the message to the remote user
is handled automatically by the protocol engine according to the
actual connection configuration. Megaco messages are not exposed
in the API.</p>
<p>On the receiving side the transport module receives the message
and forwards it to the protocol engine, which decodes it and
invokes user callback functions for each transaction. When a
user has handled its action requests, it simply returns a list
of action replies (or a message error) and the protocol engine
uses the encoding module and transport module to compose and
forward the message to the originating user.</p>
<p>The protocol stack does also handle things like automatic
sending of acknowledgements, pending transactions, re-send of
messages, supervision of connections etc.</p>
<p>In order to provide a solution for scalable implementations of
MG's and MGC's, a user may be distributed over several Erlang
nodes. One of the Erlang nodes is connected to the physical
network interface, but messages may be sent from other nodes and
the replies are automatically forwarded back to the originating
node.</p>
</section>
<section>
<title>Single node config</title>
<p>Here a system configuration with an MG and MGC residing
in one Erlang node each is outlined:</p>
<image file="single_node_config.gif">
<icaption>Single node config</icaption>
</image>
</section>
<section>
<title>Distributed config</title>
<p>In a larger system with a user (in this case an MGC)
distributed over several Erlang nodes, it looks a little bit
different. Here the encoding is performed on the originating
Erlang node (1) and the binary is forwarded to the node (2) with
the physical network interface. When the potential message reply
is received on the interface on node (2), it is decoded there
and then different actions will be taken for each transaction in
the message. The transaction reply will be forwarded in its
decoded form to the originating node (1) while the other types
of transactions will be handled locally on node (2).</p>
<p>Timers and re-send of messages will be handled on locally on
one node, that is node(1), in order to avoid unnecessary
transfer of data between the Erlang nodes.
</p>
<p></p>
<image file="distr_node_config.gif">
<icaption>Distributes node config</icaption>
</image>
</section>
<section>
<title>Message round-trip call flow</title>
<p>The typical round-trip of a message can be viewed as
follows. Firstly we view the call flow on the originating
side:</p>
<image file="call_flow.gif">
<icaption>Message Call Flow (originating side)</icaption>
</image>
<p>Then we continue with the call flow on the destination
side:</p>
<image file="call_flow_cont.gif">
<icaption>Message Call Flow (destination side)</icaption>
</image>
</section>
</chapter>
|
