Видео

Like the presentation, gives insights on practial apporches

brand name booking date together cannot ensure a unique booking key

The voice is a bit low..it's only for me :(

Very informative video!

Thank you!

Useful and great intro. Covers essentials

This was a great presentation. This part in particular is what made data vault click. There is no "right model". The right model is the one that works for the needs of your project.

Booking Details satellite should have Booking hash key but it was mentioned Customer Hash

Very useful series. Thank you!

❤

Information does not exist in this "physical realm" Only structured data. Go Datavault.🤩

Can you please help me with these . Would be great if you can explain with some examples. 1. Why link to link relationship is not recommended in RDV? 2. In BDV bridge table, if we are storing only hash keys( not natural keys), then how in fact/dimension we are going to get natural keys?

Can you please help me with these . Would be great if you can explain with some examples. 1. Why link to link relationship is not recommended in RDV? 2. In BDV bridge table, if we are storing only hash keys( not natural keys), then how in fact/dimension we are going to get natural keys?

Can you please help me with these . Would be great if you can explain with some examples. 1. Why link to link relationship is not recommended in RDV? 2. In BDV bridge table, if we are storing only hash keys( not natural keys), then how in fact/dimension we are going to get natural keys?

Around 2m37s for the link table the natural key isn't the concatenation of customer hash and booking hash? If it is so shouldn't the "customer booking hash" be calculated as the md5 of them instead of the md5 of the natural keys extracted from of the staging table? The md5 of the concatenated natural keys shouldn't happen to be identical to the md5 of the concatenated hashes of the natural keys, right?

😩 *Promosm*

Helpful. Thank you.

This is a fantastic talk from Patrick going into great detail

what's the difference between hash_diff column and other hash column at 7:42. Aren't both same?

Genius!! Thnx a lot!!

LIKE the technique discussions, but DO NOT like the political examples.

data vault what form of data?

Very good presentation and content on streaming into a data vault. Thank you!

highly informative and was latched on to it all the time. Excellent.

Very good explanation!

Thank you!

Can you explain why you do not recommend implementing a DV on a sql server? Why do you think snowflake is a better choice?

Thank you. I was interested to look into satellite loading sql as well. Which book did you ask to refer to?

Thank you! This slide was the most important slide for me. Showing the architecture enabled me to understand how Data vault works and how it can be used in my organization.

þrðmð§m 😡

Gert presentation Neil. You display a lovely style of communication. I recommend your videos to those coming to terms with DV for the first time

thank you so much

Is there a next session continuation for this video..This is really good 👍

Can you provide a link to the white paper mentioned in the beginning of the video?

I love it .. :) Always there is a simple solution to a hard problem. You are making data modeler job less ..

In my entire life I was waiting for the Puppini Bridge.

Hi Neil, I like this training a lot : ) One question though - why does it mention the effective dates in the satellites? I'm quite sure that DV 2.0 is about insert-only architecture, so getting rid of any updates, which means we cannot have the effective / end dates in the raw vault. Please comment : )

Interesting. Look like a great solution and I feel like I need to go deeper... I have a first question... In this model... What is the convenient way to implement more than one relationship to the same table? EG: Email -> to:Contact from:Contact

My concern here is that introduces a new ETL layer. You have to maintain the bridge table. What if my fact tables are very large, with several billion rows? That can be expensive to maintain. If I insert or delete from the fact table I have to do the same on the bridge. It can get problematic very quickly

40:19 A wild cat appears! Just needed to timestamp that for reference. Great video! Learned a lot!

Excellent presentation. This guy solves real world problems. I liked the way the talked about unit of work for the links.

If each stack of colors represents a dataset then the model is incorrect. The model contains just one link (red, purple and blue) and the link is coloured in Yellow. That link is Correct only if the values of the yellow fields are non hubs making the link peged leged. I am missing the link from stack2(purple, light purple, red, green and blue) and the link from stack3(yellow and blue). Why didnt you model them? This model is not Jedi safe, data Cannot be recreated from the model as it was in the sources.

Data vault! Seriously! Cheap and fast? Hmmm. Quick and easy? Hmmm.

can we have multiple satellite for a hub, if yes then do we keep natural key from both satellite in to hub?

p̶r̶o̶m̶o̶s̶m̶

nice presentations, i was wondering on last use case about the snapshots asked on demand by users, would they get loaded on independent tables or simple something like timetravel posfix on sql command at <timestamp> as feature in Snowflake ?

Before you continue to publish this blunder any further I'd recommend to read Sorting and Searching from Donald Knuth. You can't cover all and anything with hash keys. In particular not in databases. You simply do not know the hash function for arbitrary data. Hash-keys as indices in compilers are perfect, but not in databases. You will not get a structured data-model. Without defining cardinality you won't get anything useful with hash-keys.

Hi Neil. This was a very interesting session. On the DevOps slide, the diagram showed Kanban flowing into GIT and Wiki having some icon. What system are you using there?

The green hub should have 2 satellites instead of one?

I still struggling to accept Satellites using system load_date as pk instead of a more business date, just because "we dont control" source dates. In my opinion would be better to use extract date in what i call "real data date", i done this in several data lakes ingestion processes that would grab from rdbms like oracle 1 extra column on each table: instead of: SELECT * FROM <some table> do: SELECT sysdate extract_ts, * FROM <some table> then no matter if ingestion of data takes a week to get loaded and eventually loaded after more recent data, we always have proper scd2 "real data date". Dates are one of the most important business aspects, and as such I congrat you making a video calling attention on this.