Towards a harmonised Tispan Subscription model TISPAN WG 8 – 3 GPP SA#5 Joint meeting Sophia Antipolis, May 14 th - 15 th 2007. Source: Steve Orobec BT
Contents • The Challenge – part 1 • The Challenge – part 2 • TISPAN Problem Space • Bringing it together • Result • Summary © British Telecommunications plc
The Challenge – part 1 • 3 GPP’s fine for the management of mobile networks but is not rich enough to cater for next generation networks (NGN’s) • The challenge is to capture the useful aspects of 3 GPP with NGN needs – Agree consensus with both Tispan AND 3 GPP – Protect existing investment in 3 GPP – To create a whole new SOA architecture model for the management of NGN services – Do this in an agile way such that it reduces integration barriers – Can be Implemented in current industry standard tools along with industry best practice leverage the work of W 3 C, OASIS and OMG – Harmonise with other strategic standards e. g. TMF SID & MTOSI © British Telecommunications plc
The Challenge – part 2 • Service providers are looking to structure their OSS around TMF SID, MTOP & e. TOM. • Service providers are looking for a solution to fixedmobile convergence and to develop value add services • To do this we need a SOA model with a contract API between OSS e. g. CRM, Service BSS and IMS – This implies standard operations defined and understood on both sides - verbs – Data binding objects, which are defined and understood by both sides – nouns – A method to integrate these data binding objects into the existing models at either side of the interface © British Telecommunications plc
3 GPP Observations • Multiple-Inheritance (Use of Top as root class): implementable in C++, requires use of workarounds in Java and C# and not supported in XML Schema. • Cyclical dependency (Class A depends on Class B and vice-versa) – Far too many associations, reduce the number by applying ‘Law of Demeter’ i. e. each class talks only to it’s neighbour. Each association is an attribute in a class, each m-n relation is a list • Not aligned with e. TOM • 3 GPP Model uses UML 1. 5 – This UML is obsolete and non standard – current version is 2. 1 – Tolls supporting UML 1. 5 like Rational Rose no longer exist – I can’t buy it! © British Telecommunications plc
TISPAN Problem Space • Problem is that the 3 GPP model is different architecturally from the TMF model that BT and many OSS vendors are adopting and IMS will need to integrate with • Implementation detail not clear to outside integrator – e. g. ME! • Current TISPAN documents are only high level. E. g. Powerpoint • 3 GPP has a concept of an IMS but it is limited in scope – Limited support for Subscriber & multiple Users using different access technologies such as DSL and Mobile – Inheritance hierarchy problematic e. g. Service ‘is-a’ Managed Element – Inconsistent use of e. TOM layered architecture • It’s too tightly coupled – TISPAN vision is a distributed, loose coupled SOA – Each association or aggregation increases coupling – Re-structure according to Law of Demeter © British Telecommunications plc
Bringing it together • BT vision is the creation of a ‘master’ TISPAN model in OMG UML 2. x which is line with use of Industry trends taking the best of 3 GPP – this should be derivable from the TISPAN model • The short term goal is to take the best of both models along with the TMF SID • Using an Agile development driven approach develop a prototype constructed from a hybrid 3 GPP-TISPAN UML model to validate ideas • This will allow certain concepts to be proven plus technical issues with the models to be discovered and resolved • As an added bonus – it deepens the model understanding! © British Telecommunications plc
Summary • • Move the TISPAN model debate away from high level discussion to hands on UML development Promotion of UML 2. x as the technology neutral descriptive language, this supports the use of meta-models to describe SOA contracts plus automatic generation of technology specific models e. g. Java, C++, IDL etc Goal is to have TISPAN as the technology neutral and overarching model and generate the 3 GPP model as a subset – possibly in C++, alternatively to generate a full Java solution for NGN. Further goal is harmonisation with TMF – initially by mapping but in the long term by defining a multi standard meta-model from which others can be derived Use of tools such as Borland/IBM allow for rapid prototype generation to test ideas and challenge existing dogma New models and ideas can be trialed – currently prototyping SID OSS to IMS interaction – the results of which will feed back to both TISPAN and TMF Agile UML modeling offers future promise of harmonising all our Business OSS models under a single meta-model, with the ability to generate artefacts such as Java, WSDL directly from the UML © British Telecommunications plc
Скачать презентацию Towards a harmonised Tispan Subscription model TISPAN WG
595b473f2521b5fccdefd1c34d3ce33c.ppt