I have always been under the impression PRACK is for transmission over unreliable media, i.e open internet. Since most of our customer base is SIP trunking and registered devices for Broadsoft hosted PBX, where the last mile is QoS controlled for signaling and media, so PRACK should not be required in both scenarios. Most ringback scenarios have 180 without SDP arriving back at the caller, signaling local RBT for the endpoint is required, whether the RBT is generated by a device closer to the endpoint or the endpoint itself is the discussion. Broadsoft DMS has profiles for the phones that gets downloaded to them with user logon details and other options like the option to generate local RBT probably via .wav file or similar. The concern I have is that if the media comes from a centralized source like a BSFT media server, issues on the MS will have customers stranded with garbled/no RBT/wrong RBT. My employer due to a legacy TDM background prefers non-endpoint RBT generated (he would, seeing that in TDM the remote side or closest network exit to the phone generates the RBT). I have always taken it for granted that best common practice is for the device to generate its own ringtone in a SIP environment. I am now forced to question this logic, as the argument is that in TDM ringback has no issues, but can be problematic in SIP. From: John S. Robinson [mailto:jsr at communichanic.com] Sent: Tuesday, November 06, 2012 5:07 PM To: voiceops at voiceops.org; johnbotha at hotmail.com Subject: Re: [VoiceOps] Local ringback on SIP 180 from Broadsoft media server vs phone generated RBT In general, I advise my clients to not generate ringback tone on phones. Best practice is to have phones indicate alerting state by responding with 180 RINGING with no SDP, and require PRACK method. Media flow from phones before 200 OK may cause other problems, especially in a simultaneous ringing environment. One common practice is that 180 RINGING is sent without SDP, and a media gateway closer to the caller furnishes the ringback tone. PRACK is optionally used to ensure that the provisional response is reliable. Another common practice is to send 183 PROGRESS with SDP indicating the availability of early media and progress tone, but that practice isn't bullet proof. Hope this helps. If you would like additional discussion, please feel free to contact me off list. John S. Robinson jsr at communichanic.com Communichanic Consultants, Inc. On 11/5/2012 23:40, John Botha wrote: We are debating on having our Broadsoft media server generating RBT on a SIP 180 vs having the phone generate it locally, which is currently the case. We have found sporadic ringback issues when the phones generate local RBT when it plays one ring and then goes silent or no RBT at all. This only happens once out of approximately 10 calls though, and signalling looks normal (ie no 180 followed by 183 or duplicate messaging). The end-user hardware devices are mostly Polycom. Does anyone have some experience, advice and information this? _______________________________________________ VoiceOps mailing list VoiceOps at voiceops.org https://puck.nether.net/mailman/listinfo/voiceops