CODEWITHSUNDEEP
firebasegoogle-cloud-firestore
Jonas
Jonas

Reputation: 7885

Theory question: what strategy is faster? Querying a lot of documents vs query fewer documents and then load some?

I'm wondering whats the better structure for my Firestore database.

I want to create some sort of appointment manager for employee where I can show every employee its appointment for some date. I have thought of these two options:

  1. Option:

Every employee has a collection Appointments where I save all the upcoming appointments. The appointment documents would have a column date.

When I want to load all appointments for a date I would have to query all appointments by this date.

  1. Option:

Every employee has a collection Workdays with documents for each day. These workday documents would have the column date. And then a collection with Appointments where I save the appointments for a workday.

When I want to load all appointments, I would have to query the Workdays collection for the correct date and then load all its Appointments.

I expect an average workday to contain 10-20 appointments. And let's say I save appointments for the next 30 days. For option 1, I would then have to query 300-600 documents down to 10-20.

For option 2 I would have 30 documents and query it for 1 documents. Then load around 10-20 documents.

So in option 2 I would have to query fewer documents, but I would have to wait until the query is finished and then load 10-20 further documents. While for option 1, I would have to query more documents but once this query is finished I wouldn't have to load any more documents.

I'm wondering what option is the faster for my use case - any thoughts?

Upvotes: 1

Views: 67

Answers (3)

Ronnie Smith
Ronnie Smith

Reputation: 18565

Documents are not necessarily tabular (columnar). Keep it simple, follow their documentation, Choose a data structure, and do not overthink optimizing search. Leave query optimization to the Firebase platform/team as there are several search approaches which might be implemented, depending on the type of data you are querying for. Examples include source Wikipedia:

Dijkstra's algorithm, Kruskal's algorithm, the nearest neighbour algorithm, and Prim's algorithm.

Again, provided you basically follow their data structure guideline, the optimal search approach should be baked in to the Firebase/Firestore platform and may be optimized by them when possible. In short, the speed of the compute platform will amaze you. Focus on higher level tasks relating to your particular app.

Upvotes: 2

John Wu
John Wu

Reputation: 52240

The first option is rather straightforward, and is definitely the way you'd do it with a relational database. The date column could become a natural foreign key to any potential workday table, assuming one is even needed.

The second option is more complicated because there are three data cases:

  • Workday does not exist
  • Workday does exist but has no appointments in the list
  • Workday exists and has appointments

In terms of performance, they are not likely to be very different, but if there is a significant gap, I'd gamble option 1 to be more efficient.

Upvotes: 1

Doug Stevenson
Doug Stevenson

Reputation: 317467

If the total number of documents read in each case is the same, the faster option will be the one that reduces the number of round trips between the client and server. So, fewer total queries would be better. The total number of documents in the collection is not going to affect the results very much.

With Firestore, performance of queries scales with the size of the result set (total number of documents read), not with the total size of the collection.

Upvotes: 1

Related Questions