Step 11. The first call
The devices are registered. Let us make a call.
Case: user A picks up the handset, dials a known number of user B, and the call arrives at user B’s registered device. User B picks up, and a voice connection is established.
According to the scheme from step 10, the SIP INVITE request reaches SG and is proxied to B2BUA.
A distinctive feature of the INVITE request transaction is that it is stateful and controllable: in particular, preliminary responses can be sent for it, and it can be canceled with a CANCEL request. A successfully processed INVITE request results in a SIP dialog between two UAs. Repetition: when processing SIP INVITE requests for calls, B2BUA acts as the opposite UA for the request initiator. It organizes authorization checks, routing, callee lookup, and the second dialog leg, including multiple calls. For both subscribers (caller and callee), it is a full-fledged UA, thus combining two SIP dialogs within one call. The concepts of SIP dialog and B2B dialog differ.
-
SG, processing the INVITE transaction, randomly selects one of the available B2BUAs on the site. Subsequently, the entire SIP dialog of the initiating leg of this call passes through exactly these SG and B2BUA instances. Special fields in dialog-related requests serve this purpose: Route and Record-Route.
-
B2BUA, with the help of DC, performs call routing. This process occurs on the current site but may be proxied to sites serving the affected domains. Routing is an algorithmically complex process that supports, among other things, transferring responsibility to another domain. The result of routing is a list of accounts, their domains, and the rules for calling them, for example, call time and failover redirect rules. The process may run sequentially on several sites, due to routing to another domain or due to a call from a device registered in roaming.
-
In the current case, after routing B2BUA receives a result pointing to user B’s account. Using Registrar, it searches for user B’s registered devices. This process occurs on the current site but may be proxied to another site if the initiator is registered in roaming, or if user B is in another domain not served on the current site. It is worth noting that calling registered phones is far from the only possible result of the routing stage. Calls can also go to external networks and to subscriber feature codes.
-
B2BUA forms a B2B dialog, creates a second leg in it, and starts the call stage by sending a SIP INVITE request to the SG through which user B’s device was registered. If user B has several registered devices, B2BUA calls them all at once, possibly sending INVITE requests to different SGs, including those on other sites. This process is called forking; each called device is a fork, a separate SIP dialog is created for each fork, but at most one will complete successfully.
-
On one of the devices, user B picks up, and their device sends a 200 OK response, while the others receive a call rejection via the SIP CANCEL request. Then, according to the protocol and the scheme:
Consequence 1: When one user calls another, there are always exactly 3 SIP nodes in the chain between the two SIP devices: SG, B2B, SG. If they are registered through the same SG instance, it is the same SG serving two legs, that is, two SIP dialogs.
Consequence 2: The caller’s SG and B2BUA are always located on the same site, while the callee’s SG can be on any site.
Consequence 3: B2BUA creates B2B dialogs only on external initiative; thus, any B2B dialog has an initiator and a call recipient. However, there is one exception: the call pickup function, where two initiators are connected in one dialog. Subscriber A is the one who initiated the original call, and subscriber B is the one who dialed the call pickup feature code.
Audio and video data associated with the call are transmitted over the RTP protocol. During the connection establishment transaction, parties A and B exchange media stream descriptors over the SDP protocol (wrapped in the body of the INVITE request and the 200 OK response). Media traffic is handled by the media server within the MG role. The performance of this function is very critical, and it is implemented in a separate RTX application that is part of the platform. In essence, the MG role is a wrapper over the RTX media server.
MG (through the RTX media server) exchanges traffic, mixes sound in conferences, and performs transcoding: repackaging traffic from one codec to another (audio or video). One average server can serve 200-500 simultaneous calls. The exact number depends on many factors: CPU performance, the server's involvement in tasks other than media server tasks, the presence of transcoding in calls, and the presence of video traffic and video transcoding in calls. Thus, expanding the number of servers allocated to media servers on a site grows along with the requirements for the number of simultaneously served calls. They are made redundant and scaled in Active-Active mode.
RTX is controlled via the MEGACO protocol. The opposite side implementing the MEGACO protocol is the MGC role. Its main responsibilities are selecting the freest suitable media server from its group and translating control of call media contexts: creating a media context, adding media terminations to it, setting the media topology, termination properties, and starting playback and recording. Several MGC groups can be configured on a site, each made redundant in Active-Passive mode and managing several MGs.
The media context for a call is created on the initiative of B2BUA, which selects an MGC group immediately after creating the B2B dialog. If the selected group fails, B2BUA tries another group. Then each new SIP message at the signaling level of controlling one side’s SIP dialog leads to adjustment of the media context: media terminations are added, removed, or modified. If media gateways are also used by other SIP servers serving the call, each of them manages its own media context.
1. Conversation between two internal subscribers. The simplest case. |
2. Conversation between an internal subscriber and IVR |
3. Conversation between an external subscriber and IVR |
4. Conversation between an external subscriber and IVR. The case with a border media gateway. |
Consequence 4: Call traffic is exchanged by the media server of the site where the B2BUA instance serving the call is located.
Consequence 5: However widely the sites are spread, traffic always takes the shortest route through the site to which the call initiator’s device is connected.
Consequence 6: For calls between two sites to work, direct network visibility must be ensured between all B2BUA, SG, and MG instances of those sites.
| Term | Definition |
|---|---|
|
! |
|
! |
|
! |
|
! |
|
! |
|
! |
|
! |
|
! |
|
! |
|
! |
|
! |
|
! |
|
! |
|
! |
|
! |
-
Next step: Step 12. A call from one domain to another
-
Previous step: Step 10. Connecting a SIP phone