关系数据库与非关系数据库-运动结果和架构

Question

I'm looking to consolidate a number of historical sporting results in a database which will allow me to query them for the future purpose of building apps and visualisations.

My understanding of this area leads me to believe I could be taking a non-relational approach (Such as Amazon's DynamoDB) so that I don't have to take database resource, size etc into consideration as the size of the data, and the querying of it, grows.

I then have plans to use Amazon Lambda to handle the import of data and handle responses to incoming queries.

My two questions are:

• Is this valid approach?
• What would a non-relational schema look like?

To help context the second question, if this was a relational database the tables would look something like this.

• Events
• Races (Where a result and associated fields are stored)
• Athletes
• Teams

Answer 1

If your query will have lots of joining to those tables, relational db will be preferred. Although it's possible to use DynamoDB, you'll have a hard time in joining those tables. The feature of DynamoDB that can make it hard for developers is the indexing. you have to search with the hash key. So you have to be careful in choosing the index.

However, if the data is just like a lookup, eg race ID 123, winner is Athlete A. DynamoDB would thrive as there are advantages of using it (scalability, minimal management, etc)

Personally, I don't like dynamoDB when listing something (using search, pagination, etc) You can build a search index of course but for simpler application, it would be an overkill.

ALTERNATIVE

if you are only interested in building the visualization and aggregation of the data, you can use redshift or S3 to store the data. And run some aggregation function later on. Then save the aggregation into the dynamodb or rds.

I consider dynamodb and RDS databases for data which are queried by end users only and use.