CODEWITHSUNDEEP
amazon-web-servicesaws-lambdaamazon-dynamodbalexa-skills-kit
zuba
zuba

Reputation: 395

Search dynamoDB using more than one attribute

I've created a skill with the help of a few people on this site.

I have a database and what I want to do is ask Alexa to recall data from my database. I.e. by asking for films from a certain date

The issue im having at the moment is I have defined my partition key and it works correctly for one of my items in my table and will read the message for that specific key, but anything else i search it gives me the same response as the one item that works. Any ideas on how to overcome this?

Here is how i have defined my table:

let handleCinemaIntent = (context, callback) => {    
  let params = {
    TableName: "cinema",
    Key: {
        date: "2018-01-04",
    }
  };

Just as a side note, I will have the same date repeating in my partition key and from what I understand, the partition key needs to be unique; so i'd need to overcome this.

Upvotes: 0

Views: 358

Answers (1)

Mike Dinescu
Mike Dinescu

Reputation: 55750

You have a few options for structuring your DynamoDB table but I think the most straightforward is the following:

You can set up your table with a partition key of "date" (like you have now), but also with a sort key which would be the film name, or some other identifier. This way, you can have all films for a particular date under one partition key and query them using a Query operation (as opposed to the GetItem that you've been using). You won't be able to modify the existing table to add a sort key though, so you will have to delete the existing table and recreate it with the different schema.

Since there is generally a rather limited number of films for each day, this partition scheme should work really well, assuming you always just query by day. Where this breaks down is if you need to search by just film name (ie. "give me the dates when this film will run"). If you need the latter, then you could create a GSI where the primary key is the film name, and the range key is the date.

However, you should pause a moment and consider whether DynamoDB is the right database for your needs. I say this because Dynamo is really good at access patterns where you know exactly what you are searching for and you need to be able to scale horizontally. Whereas your use case is more of a fuzzy search.

As an alternative to Dynamo you might consider setting up an ElasticSearch cluster and throwing your film data in it. Then you can very trivially run queries like "what films will run on this day", or "what days will this film run", or "what films will run this week", or "what action movies are coming this spring", "what animation films are playing today", "what movies are playing near me"

Upvotes: 2

Related Questions