Team,

Allow me to kick off this thread by quoting TR-139 (p. 23) regarding the scope, and highlight keywords in blue: "An operation phase concomitant with execution where services are managed according to QoS, charged according to usage apportioned according to sharing rules if service components in the blend belong to third parties. This phase also includes reporting on service quality, usage correlated with underlying network resource capabilities usage to provide useful information on the cost, revenue and quality; patterns of customer behavior can be also identified during this phase."

Best regards,

Paul Lin

This topic raises many questions ...

1. what is the meaning of QoS in next generation of services? When I download a ringtone or I browse a mobile content portal I am charged, right? Is there any spec out there telling me how the QoS of such services is defined, which KPIs and KQIs to measure and compute? Or is it more like "you got it, you pay, you didn't ... we'll figure something out".  But with the ringtone, when do I know that I will never get it? Or how does my service provider know that my data volume consumption decreased over the last month because I just gave up on how slow my mobile browser was, or was it slow or just me being impatient? So what is the Quality of Service?

2. and whose responsibility is it anyway? what if the culprit is the content provider, what are the means  to detect early, constrain and compensate? Beyond B2B agreements and SLAs which we know are just soft ways to control, how can they become actionable at runtime?

3. network resources capabilities usage to deliver the service, can we control that? we can report on it ... eventually, meaning sometimes applications are hidden behind network adapters with no means for a network element to know exactly who was using it, the "Disney Little Mermaid Wallpaper" downloads or "My Best Friend For Ever" group? Subscriber's demand may peak on one or on the other and some control must exist as to how much each of these two applications can eat out of a common resource without penalizing other applications and subscribers. And what happens if all of them are on the same stadium? Remember that? The last mile ... hard to anticipate ... We talk about e2e control of resource allocation and then one day a third party is so successful or the whole demand moves in a tiny area that our control means fail! What type of solutions do we have here?

4. data in how many repositories must come together in an OSS/BSS environment to answer the question: how much does this service quality cost me and how much revenue do I get out of it? Now! A first step is of course knowing what QoS means..

QoS etc

I am uncertain about how QoS is to be modeled in SDF.

In my model, QoS of service aspects are taking place at two levels:

1)      When a subscription or agreement specifies a SLA, QoS, Policy it is to be provisioned into the run time to ensure that run time or network components can enforce the terms or alert in case of specific events

2)      The OSS should be monitoring the performances of the system for activity monitoring, reporting, auditing + to detect trouble so that OSS/BSS can resolve

On the other hand, I would not support at all the view that further is to be modeled by SDF. Again we need to have enablers / SDF services or interfaces to provision SLAs, Policies QoS settings and to monitor / measure performances. Beyond that how enforcement or measurement is done is out of scope of our work IMHO.

May be QoS might not be modelled in SDF. But today many of the Telcos especially in APAC are looking upon SDF to enhance the customer experience by addressing QoE and QoS. Given this is a hot need in the market, my opinion would be:

1. Create a QoS service in the service layer which uses the OSS platform

2. This service will need to ensure the various aspects like download success, Quality of content, billing at the right time, proper provision of content in the appropriate platform

3. Yes OSS will monitor the performance, but given the fact that SDP is expected to "Throttle" the services across layers, we might need to think of a standard / way on we can throttle the avaiabilty of services to consumers / other players in the Telco ecosystem

Hi to All Contributors on this thread:-))

Thank you for the thoughtful emails and stimulating questions. It appears TR-139 authors have a "vague" model in mind about service quality and we'll need to contribute more to paint a clearer picture.