One Schema or multiple Schemas ?

Question

I am doing project for a company which consists of 3 entities.

Data Mart would ingest to CDP 4 datasets:

- data for entity 1

- data for entity 2

- data for entity 3

- combined customer view across all entities

Would you make ona schema or 4 schemas? Is it some max number of fields which schema

arpan-garg · Accepted Answer

Hi @michael_soprano -

I’d go with one schema if all three entities have the same structure (same customer fields, same meaning). Just add a field like entity_id to tell them apart. That way, you can easily build a unified profile view and keep things simple.

If each entity’s data is very different, then make separate schemas, but still try to use the same XDM class so they can roll up into the union profile.

Also, for future proof solution and performance, Adobe recommends keeping it to max 20 datasets for Individual Profile and 20 for Experience Events, so it’s better not to over-split your data unnecessarily.

Thanks,

Arpan

KevinFosterDX · Answer

I typically combine into one dataset when:

the data is of the same class: profile, event, or custom
the data will have the same primary identity
the data is sourced from the same system
the data will be sourced the same way: streaming vs batch

If you combine profile-class data into one dataset when all of that isn't true then you may end up forced to use "upserts" (partial record ingestion) with all of the limitations that comes with that.

Sign up

Login with SSO

Login to the community

Login with SSO