I have a Countries collection and inside this collection, There are 195 countries, And the admin can add or delete any country.

In my android app, I want to know how many countries are currently there without reading all countries again because that will cost me 195 read for each user, So I decided to create a counter for this but the question is where should I put the counter? Should I create a new collection for just that counter?

CodePudding user response：

There is no singular correct answer for this, so I'll illustrate some patterns I've seen used instead.

The distributed counter example in the documentation stores the counter documents in a top-level shards collection. That's a good starting point, although I'd probably name it something like counters, metadata or counters.

In the Cloud Functions example of a child node counter for Realtime Database, we store the counter as a sibling property of the node that holds the children. So you have likes (which hold the actual likes) and likes_count, which is the counter. In Firestore that structure could also work if the count isn't global, but you're counting a subcollection of a specific document.

Finally I did some experimenting with geo-indexes myself a while ago, and there I put my counter documents in the same collection as the documents they counted, but then with a predictable name. The documents had prefix-based IDs, so I ended up with count documents geodocs-2 counts, geodocs-3 counts, etc.

Each of these is a good practice for the use-case in question, and can likely be applied to many other use-case as well. But I'm guessing there are many more reasonable, workable strategies for other cases too.