DynamoDB - how to query by something that is not the primary key

user_3648701 0
Viewed 246

So, I have a table on DybamoDB with this structure:

- userId as the primarykey (it's a uuid)
- email
- hashedPassword

I want to, as someone is signing up, find out if there's already someone using that email.

This should be easy but, as far as I know, you can't query on DynamoDB unless you are using the primary key as parameters or the sort key (and I'm not sure if it would make sense to make email a sort key).

The other way I found out was using a Global Secondary Index, which is pretty much an index table you create using another field as the primary sort of, but this is billable and since I'm still developing and testing I did not want to have expenses.

Does anyone have another option? Or am I wrong and there's another way to do it?

amazon-dynamodb aws-lambda amazon-web-services
3 Answers

Like other answers, I also think that GSI is the best option here.

But I would like to also add that since search capabilities of DynamoDB are very limited, it is not uncommon to use DynamoDB with something else for that very purpose. One such use case is described in the AWS blog:

The main querying capabilities of DynamoDB are centered around lookups using a primary key. However, there are certain times where richer querying capabilities are required. Indexing the content of your DynamoDB tables with a search engine such as Elasticsearch would allow for full-text search.

Obviously, I don't recommend using ES over GSI in your scenario. But it is worth knowing that DynamoDB can be, and is often, used with other services to extend its search capabilities.

user_2488230

Even you put email as sort key alongside of userId as primary key, you can't query only using email(unless it is scan operation). You don't want to use scan to see whether email exists in your table. It's like iterating the each value by scanning the whole table.

I think your best option is global secondary index. Another option would be creating a new table which only includes email values, but in that case you have to write/maintain to multiple tables which is unnecessary.

user_21889220

The other way I found out was using a Global Secondary Index, which is pretty much a index table you create using another field as the primary sort of, but this is billable and since I'm still developing and testing I did not want to have expenses.

As @Ersoy has said, GSI is the legit solution, even it will increase the consumed writes units.

Dynamodb is cheap for a low-traffic app and/or a test environment, but to hold these expenses flat, you can:

  1. Use dynamodb local during local devs/tests and CI builds

  2. Choose a provisioned capacity mode for your table (you may find its free-tier interesting)

user_47166850
Related