Question

I have a question on designing index in MongoDB.

Say i have a users collection and groups collection.

user {
name : "" ,
age :  19
}

group {
name : ""
members : [],
posts :  [ { date : "" , author : "" , topic : "" }, { date : "" ,
author : "" , topic : "" } ......]
}

There can be 1000’s of groups and each group can have millions of
posts. Operations I frequently perform is:

1. getting posts based on date (70%)
2. updating posts (30%)

So, essentially I need to index on date.

My question is:

Should i create a new posts collection like

posts {
name : "",  date : "" , author : "" , topic : ""
}

and create a single-value index on date in posts collection

( db.posts.ensureIndex({posts : 1}) )

OR

Should i include posts inside of group object and create an embedded
index like db.groups.ensureIndex({ posts.date : 1})

Which one is more efficient ? whats the best practice if this needs to
scale to millions of posts ?

Thanks

Answer 1

@Z5h, I think u mis-understood the problem.
The problem was getting posts of a particular group between a date range.
and storing them most effectively.

and after some thinking and research, this’s what i found out.

• First, there’s a limit on size of document ( currently 16MB ), and as post schema/posts increase in size, this may stop scaling some day as number of posts increase.and u cannot add an index to search within array of sub documents, as indexes are only across collections.

• Second, If posts are stored as embedded sub documents, there would be no way to search within posts for a group in a date range. I have to get entire posts array and do processing on client side which is inefficient. There’s no way to compare array objects based on a field in sub document as of now. refer this

• Hence better way is to create a separate posts collection , and have foll data

  posts{
  group_name : objectID(<ID in groups collection>),
  date : "",
  author : "",
  topic : "" 
  }
  

By this way, i can as well create index on date and get all data for a group in a date range more effectively.