questions to the modeling of TAPI

[YuChaode]

A lot of thanks to Nigel for the sharing of TAPI modeling. After i read the TAPI modeling slide, i have got some questions for it. And if the other friends also have some other puzzles, please also raise here.

My questions are:

1. May I consider the device object is equivalent to our Network Element? If it is, how to describe the relationship between NE and rack?
2. Is Chassis an equipment or a holder? According to TMF, chassis is a holder.
3. I think strand joint is a passive equipment, how can the controller discover passive equipment? This could be helpful for the modeling of optical splitter in Access domain.
4. Is physical span the fibre or cable object? And in Optical domain, fibre is usually a pair with two direction, is physical span one of the directions?
5. What is abstract strand inside the physical span? If a picture can be provided, it will be perfect.
6. In the physical span, there are connectors, strand joint abstract strand and some more detailed of objects. What is the scenario to use these object?
7. Are the holes on board called port, even if there is not a SFP inside? SFP is the equipment supported by port? What is the relationship between access port and SFP？

some picture from my owned understanding: Questions to TAPI modeling.pptx

[nigel-r-davis]

Considering your questions (to reduce delay I will deal with each question in individual comments).

Question 1: May I consider the device object is equivalent to our Network Element? If it is, how to describe the relationship between NE and rack?

Response: In ONF Core we have dismantled the concept of NE (and Managed Element) into several separate concerns (management-control, functional, physical..). We need to be careful with Device in that context. Device is a collection of physical items that are grouped from a management-control perspective. Management-control focusses on access to the information, but Device does not necessarily reflect the management-control access that was actually achieved by the controller as it may have been assembled from several distinct accesses to the equipment that have been consolidated into a single "thing" by the controller that is then reporting through TAPI. I think it may be equivalent to a view of the Network Element (we would need to discuss this in detail). So to the second part of your question. Device does not related to rack. The equipments collected in the device each relate to rack (and, as per our discussions on the call, some may be rack). Device is identified in the TAPI model diagram color scheme (your slide 3) as a logical grouping of physical things.

[nigel-r-davis]

Considering your questions (to reduce delay I will deal with each question in individual comments).

Question 2: Is Chassis an equipment or a holder? According to TMF, chassis is a holder.

Response: The diagram from TMF (your slide 1 (coincidentally one of my diagrams from my TMF days :) ) is a representation of the MTNM model as it was in ~2004. That was derived from older thinking and previous models. We became increasingly uncomfortable with the model that led to the "solids and spaces" analysis that a few of us did ~2006 (I may be wrong with the timing although I have a very strong memory of one of the meetings!). That led to us understanding a correction to the model, but it was not until we were doing the ONF Core work in ~2014 that we started to bring that insight to the model. The "solids and spaces" thinking taught us that the chassis, shelf and rack should have been an equipment that had holder positions and should not have been a holder. The TAPI model reflects this. There was also the notion of zones which was discussed but never incorporated. This allows a large space to be divided into smaller (potentially overlapping) spaces (using rules that are not directly derived from any physical demarcation) and for a solid to similarly divided. We may want to consider this in conjunction with the solids/spaces clarity. I should also note here that the "address" ( the labeling of a position) of an item often has a zonal aspect to it. I should also note that our focus is field replaceable units and hence we need not concern ourselves with many of the complexities (for example, whilst the solids/spaces/zones approach can be used to define and represent the space taken by and pin positions of a surface mount component, this is not relevant for our problem). We also do not allow holders in holders. A space is defined within a solid and a solid can be placed in that space etc. So the model is strictly equipments contain holders and equipments can be placed in holders. A holder can never be stand alone. In the model the holders are restricted spaces that can take at most one equipment. A final important note is that the location of an equipment (its "address") is in terms of holder locations and any zones. The equipment does not feature in the address because of the 1:1 between the equipment and the holder that holds it. If there could be many equipments in a holder then the equipment would also feature in the address. It was this that essentially led to the confusion in the MTNM model that we eventually understood and corrected.

[nigel-r-davis]

Considering your questions (to reduce delay I will deal with each question in individual comments).

Question 3: I think strand joint is a passive equipment, how can the controller discover passive equipment? This could be helpful for the modeling of optical splitter in Access domain.

Response: You are absolutely correct that strand joint is passive as is the actual strand. Before I get on to discoverability, the intention of abstract strand and strand joint was to allow the details of the fiber and joint impairments to be allocated (if known/relevant). During the presentation I noted that the model is extremely flexible in this area. The abstract strand can summarize the effect of many serial strand segments and strand joint can also be used to summarize or detail properties. Some of the information on the strand and the join may generated during planning, other information may be gathered during the installation. Actual measurements can be taken using tools such as an OTDR which can measure the effect of the joints (reflections etc.). The OTDR may be used during installation, but some equipments offer integrated OTDRs. Where there is an integrated OTDR it is possible to measure the effects of the joints in an overall physical span. The abstract strand and strand joint may be more relevant in a tool that abstracts the effect of various fiber properties to the overall physical span that are then used in viability calculations. It is likely that in most cases the network devices will not need to represent the strand joints in any way (the exception is where the device has an integrated OTDR where some representation may be valuable). Depending upon the controllers role it may need some level of representation.

[nigel-r-davis]

Considering your questions (to reduce delay I will deal with each question in individual comments).

Question 4: Is physical span the fibre or cable object? And in Optical domain, fibre is usually a pair with two direction, is physical span one of the directions?

Response: Physical span represents the adjacency between an access port on one device and an access port on another device. It can be a grouping of one or more abstract strands. The abstract strands may relate to only one direction of flow or two directions of flow where the physical span is bidirectional. The strands represent segments of fiber where each abstract strand may represent one or more fibers in a chain. The physical span does not represent a cable (other than in an extremely degenerate case where the two adjacent ports are connected by one continuous cable where that cable only has two strands). In most cases, the physical span represents parts of many cables. Interestingly, it would be possible to use the physical span to represent a cable if all the strands in the cable were represented individually be abstract strands, but that was not the intention.

[nigel-r-davis]

Considering your questions (to reduce delay I will deal with each question in individual comments).

Question 5: What is abstract strand inside the physical span? If a picture can be provided, it will be perfect.

There was a figure in your slide pack that you extracted from TR-547 that shows abstract abstract strand. The idea behind this entity is:

• it is necessary to represent impairments etc. of the adjacency between two device.
• impairments etc. are meaningful on a per fiber strand basis
• sometime the end-end accumulation of impairments etc. through multiple serial fiber strands is sufficient, sometimes it is necessary/desired to detail each individual strand between splices as well as the splices etc. and sometimes something between these two extremes is required. The abstract strand can represent a single fiber strand or a chains of strands where that chain of strands may be the whole span between two devices of part of some span. The abstract strand can be simply identified as adjacent to another abstract strand or it can be connected to another strand via a joint. This model provides the flexibility to show various levels of detail from the most abstract to the most detailed per strand & join.

[nigel-r-davis]

Considering your questions (to reduce delay I will deal with each question in individual comments).

Question 6: In the physical span, there are connectors, strand joint abstract strand and some more detailed of objects. What is the scenario to use these object?

Hopefully, my earlier answer helps on purpose and scenarios for abstract strand and strand joint. The model does not explicitly represent connectors as explicit entities, the connector detail is only conveyed via data in other entities.

If this is not sufficiently clear, please ask further questions.

[nigel-r-davis]

Considering your questions (to reduce delay I will deal with each question in individual comments).

Question 7: Are the holes on board called port, even if there is not a SFP inside? SFP is the equipment supported by port? What is the relationship between access port and SFP？

Where an equipment has a "hole" for insertion of another equipment, then the "hole" is called a holder. The holder is always present even when there is no equipment present in the holder.