relationship between one hub and satellite of another hub

Hello All,
In data vault design, I’m facing the relationship between one hub and the satellite of another hub. Is it possible that this situation shows that there is some problem in the model?

Please let me know your idea.
Best Regards,

why don’t you have links?

1 Like

Thanks for your response.

As someone new to the DV area, I’m currently tasked with designing the data vault model for a manufacturing company.
Based on the books I read, it is impossible to establish a connection between a hub and a satellite from another hub, and each satellite belongs to only one hub and remains like a leaf that does not continue.
I have transitive dependencies Hub A linking Hub B, Hub B linking Hub C, and Hub C linking Hub A. Each link has its own satellite; they belong to one unit of work, which is called the process CC.
Also, Hub A, Hub B, and Hub C have another relationship, called link pre-process, which is also one part of the UoW.
The link pre-process has a relationship with the final output of the UoW. I assumed that the final production is Hub P, and the pre-process has a 1-to-M relationship with Hub P. Then, it should be a link between the link pre-process and Hub P.
On the other side, Hub P has a relationship with Hub PL ( 1 to M relation from Hub PL to Hub P).

If I were to modify this design and make Hub P a Satellite of the Link pre-process (1 to M), would it be possible to establish a link between the Satellite pre-process and Hub PL?

Mate there is suite of problems of not understanding link tables to begin with.

From the context you have given it is not clear if you are talking about Ensemble DV or DV2.0; or even if you will be incorporating best practices and techniques that will make your model work. The DV is supposed to be representative of your business ; it’s why Dan Linstedt used the words “Enterprise Vision”, do not build a data vault for the sake of building a data vault — you will be building a new legacy platform. You might be better off sticking to traditional star schemas.

However, if you do wish to pursue a DV, please put yourself in the reader’s shoes. How would anyone make sense of what you have put here. You can see in some of the posts on this forum made visual representations of the model they’re proposing. I suggest you do the same.

1 Like

Thank you so much for your consideration.
You are 100% right that my explanation was confusing, and I’m sorry about that.
My goal in having a DV is to integrate different enterprise sources into EDW and use them in the data mart.
The business process for one resource type starts from the right side of the image. I want to integrate different sources of type A into EDW via the DV model. We have many kinds of sources, so I still need to add them to the model, and I am currently only focused on type A.
You can find the visual representation in the attached image.

My main problems are links on links, which are not recommended in many references, and links between links and hubs (like a triangle).
If I want a link between Hub_C, Hub_H, and Hub_S as Link_C_H_S, then again, I need a link between Link_C_H_S and Hub_PC. I hope my problem is clear to the readers.

Thank you so much again for your attention,
Best Regards

the very concept of L2L constructs does not make sense; think about it. What participates in a link? hubs and their business keys; how could another link table relate to that link — via one of the hub tables which effectively means both share a common hub table — therefore the relationship is link to hub to link.

Link tables are the recording of fact that business objects participated in an interaction, transaction or relationship. If there is nothing in the source depicting the relationship you need then how could you possibly build a link table as such?

If you need to depict a relationship as the business sees it and not how the source has provided it then here you are talking about a business vault, i.e. you must process the raw data through an idempotent business rule to produce that BV-link; we have done many cases like this before. Example SAP used in the home-loan domain applies an industry model to how facilities and the types of loan accounts relate. Unfortunately in Australia (at this particular customer) the business sees the relationship differently. We of course mapped the SAP-sourced data into RV-links but developed an idempotent business rule to take those raw vault links to populate a bv-link.

1 Like

First of all, thank you so much for your response and guidance :slight_smile:

As you rightly pointed out, the key issue in this design is depicting the L2L relationship. It’s crucial to represent this relationship as the business perceives it rather than how it’s provided by the operational source system. Business analysts argue that the L2L relationship should be the pre-product business entity (Hub pre-product in DV), which is currently missing in the operational source system.
However, this pre-product concept is not separate from Hub_C, Hub_S, and Hub_H.
It keeps the relationship between Hub_C, Hub_S, and Hub_H, which can produce the product business entity (Hub_PC).
As you rightly suggest, I’m open to considering a preproduct concept in the business vault.
Alternatively, we could introduce a new business entity (Hub in DV) that represents the pre-product concept missing from the source system. This new hub would not only fill the existing gap but also enhance the design by extending its relationship with Hub_PC, potentially improving the overall functionality and efficiency of the system.

1 Like
  1. As Patrick pointed out, there is no such concept as a link-to-link in DV 2.0 – links connect hubs only.

  2. In order to integrate data from multiple sources into a data warehouse, you will need to first identify the business key (not the source system key or surrogate key). Any links will capture a relationship between two hubs (two business keys).

  3. You said that the pre-product business entity is missing from the operational source system. If that’s the case, how are you populating the pre-product hub? You can’t have a hub in the RDV that isn’t populated from a source system.

– In Chapter 5.2.1 of Building a scalable warehouse with data vault 2.0, there is an example of an airport hub where the link is the relationship between the originating airport and the destination airport. (Something similar may be manufacturing something that is made up of multiple components that are made of other components by the manufacturing process, the individual components, intermediate components, and final product could be represented in the same hub by the same business key of item number.)

This section shows an example of trying to model a diverted flight as a link between 2 airports as well as the original flight link. In the subsequent discussion and diagram, you’ll see that the diverted flight link is still a link between the airport hubs but also the flight number hub rather than the original flight link (this is much clearer in the book).

I hope this is helpful – I’m not really clear on what is being captured in the operational system. Is it a workflow/series of process steps?


Thank you so much for your clear and practical reply.

It was really helpful. As you mentioned, I got the idea from Chapter 5.2.1 to abstain from using link-to-link and modeling as denormalization.
I added a new hub as a pre-product in BV, which is populating from the relationship among Hub_C, S, and H. Also, a BV link exists between this new Hub (Hub_Preproduct) and Hub_ PC.

I can handle an Information mart based on this model (RV+BV=DV). I hope this is within the DV concept.

Again, thanks for your help.


Hub_Preproduct ← does this mean this is a business object with its own business key?

1 Like

Thanks for your attention @patrickcuba

Business analyst mentioned that, this is one of the main business object which we have in the business process but it is missed in source system. The business key of HubPreProduct is combination of BK from RV Hubs and another one attribute → BK_Hub_C+BK_Hub_S+BK_Hub_H+ Attribute 1= BK_HubPreProduct

Then as you mentioned before, I put the Hub_PreProduct in BV
Hopefully this is correct :slight_smile:

thats not a business key mate ~ and there is no such thing as a BV hub (that is what I said - I would never propose a BV hub).

If you wanted to to pull these things into a mart together then go for it — what you’re proposing is not a DV pattern.

1 Like

You are 100 % right. Thank you so much for your input and help.

Best Regards

Dear All,

Thanks for considering my issues. I want to share my solution with you if you face the same problem in the future.
If we are facing a business relationship that is not defined in the source system, using the Exploration links is the answer to solve this issue. For reference you can read the 5.2.7 of Building a scalable warehouse with data vault 2.0 Figure 5.21.

“The three hubs (HubAirplane, HubEngine, and HubManufacturer) are interconnected to each other by two links (LinkAirplanePart, LinkManufacturer). The business might decide to analyze the relationship between those three hubs by adding LinkAirplanePartsWithManufacturer, which is a denormalized version of both links.”

Best Regards,