BackToBackCall可以支持集群UCMA应用服务器吗？ [英] Can BackToBackCall support clustered UCMA application servers?

问题描述

我们有一个建立BackToBack调用的UCMA应用程序，我们想要管理BackToBackCall无状态，以便在"拥有"的应用程序服务器中进行管理。由于崩溃/停电导致B2B呼叫失败，另一个应用服务器
可以接管该B2B呼叫的所有权。这里的场景是：

We have a UCMA app that establishes BackToBack calls, and we'd like to make the management of a BackToBackCall stateless so that if the application server that "owns" a B2B call fails due to a crash/power outage that another application server can take over ownership of that B2B call. The scenario here would be:

1。 UCMA应用程序检测到应为会议进行BackToBack呼叫。它建立B2B呼叫，并在共享存储（DB，分布式缓存等）中注册它正在管理此会议的B2B呼叫这一事实。

1. The UCMA app detects that a BackToBack call should be made for a conference. It establishes the B2B call, and registers the fact that it's managing the B2B call for this conference in shared storage (DB, distributed cache, etc.).

2。无论出于何种原因，UCMA应用程序都会死亡。

2. For whatever reason, the UCMA app dies.

3。通过故障转移检测机制（超出此线程的范围），池中的其他UCMA应用程序检测到该实例已失败并开始接收该实例拥有的B2B调用。

3. Through a failover detection mechanism (beyond the scope of this thread), the other UCMA apps in the pool detect that that instance has failed and begin to pick up the B2B calls that were owned by that instance.

4。接管B2B呼叫所有权的剩余服务器然后可以继续控制B2B呼叫而不会对最终用户造成明显中断。  

4. The remaining server that has taken over ownership of the B2B call can then resume controlling the B2B call with no apparent interruption to the end user.  

我似乎无法在文档中找到任何表明可行的内容。我只是运气不好？

I can't seem to find anything in the docs that would suggest its possible. Am I just out of luck?

推荐答案

我不认为你所描述的是可能的。在B2B呼叫中，呼叫的所有SIP流量都通过UCMA应用程序（充当背对背用户代理）传递。如果该应用服务器发生故障，两个远程端点无法互相重建通信
- 从他们的角度来看，他们正在通话的端点已经消失，即使媒体可以继续流动，因为它正在进行直接在两个远程端点之间。

I don't think what you're describing is possible. In a B2B call, all of the SIP traffic for a call is passed through the UCMA application (which acts as the back-to-back user agent). If that app server fails, the two remote endpoints have no way of re-establishing communication with each other - from their perspective, the endpoint they were talking to is gone, even though the media can continue flowing because it is going directly between the two remote endpoints.

你可以（有点）拥有一个UCMA应用程序的另一个实例"接管"使用呼叫替换（INVITE with Replaces）进行背对背呼叫，您可以在管理B2B呼叫的实例未能查看
是否有效之后立即测试，但我怀疑它是否可靠。可能值得一试，具体取决于你需要这个功能的程度。

You CAN (sort of) have another instance of a UCMA application "take over" a back-to-back call using call replacement (INVITE with Replaces), and you could test doing this right after the instance managing the B2B call has failed to see if that works, but I doubt it would be reliable. Might be worth a try depending on how badly you need this functionality.

如果你需要更多细节，请告诉我。

Let me know if you need more details.

这篇关于BackToBackCall可以支持集群UCMA应用服务器吗？的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持IT屋！