Introduction 

One common use case is that of a user running a relational database with multiple tables. They would like to create copies of those tables in a data warehouse like BigQuery in a single, simple operation. All existing data should be copied first, then new changes (inserts, updates, deletes) that are applied to the relational db tables should be reflected in the BigQuery tables within minutes. Newly created tables in the relational db should automatically appear in BigQuery. Tables that are deleted in the relational db should be delete in BigQuery. Compatible schema changes should also be reflected.

Pipelines are usually not suitable for these types of use cases, which more closely resemble replication than incremental loads. It is possible to incrementally load data from a single table to a single BigQuery table if the table never has deletes or updates and has a monotonically increasing column. Most users do not have a write pattern like this, so a better solution is required. 

Goals

Design a way for users to easily create a copy of their existing data that is continuously updated. 

User Stories 

1. Breakdown of User-Stories 
2. User Story #1
3. User Story #2
4. User Story #3

Design

Cover details on assumptions made, design alternatives considered, high level design

Approach

Approach #1

Approach #2

API changes

New Programmatic APIs

New Java APIs introduced (both user facing and internal)

Deprecated Programmatic APIs

New REST APIs

Deprecated REST API

CLI Impact or Changes

• Impact #1
• Impact #2
• Impact #3

UI Impact or Changes

• Impact #1
• Impact #2
• Impact #3

Security Impact 

What's the impact on Authorization and how does the design take care of this aspect

Impact on Infrastructure Outages 

System behavior (if applicable - document impact on downstream [ YARN, HBase etc ] component failures) and how does the design take care of these aspect

Test Scenarios

Releases

Release X.Y.Z

Release X.Y.Z

Related Work

• Work #1
• Work #2
• Work #3

Future work