Missing a most important part in my opinion, store data using open standards in an accessible platform to enable rather than anticipate evolution. Options would be e.g. Postgres, Parquet, Avro, Json, even csv. Storage is the foundational, absolutely infrastructural data infrastructure. No one cares if data pipelines infrastructure changes, but if it cannot be done, just because your data is hosed into a vendor locked-in platform, then that is the infrastructure failure you did not want in your conscience.