Search code examples
meteormeteor-publications

When should I create another publication, performance-wise?

I have one publication with 10 fields for my first template.

I have another template where I need only 5 fields of the same publication. Since user can access both, depending on the loaded template, security is out of concern.

The question:

Performance-wise, should I make another publication with only the 5 fields I need, or should I rely on my first publication?

I expect it to be a trade-off between:

  1. the amount of data sent
  2. the server side load (each of these publications would use cursor.observe())
  3. the number of users currently subscribing (the mergeBox load)

Another solution could be to create 2 publication: the basic with 5 field and another one with the 5 other fields. In my first template, I subscribe to both, in the second I subscribe to the first only.

I understand the theory, but I can't infer from it what is the best approach. I expect there are good practices in these kind of cases.

All could be equivalent too, and it would make it an unnecessary micro-optimization (and that would answer my question).

Thanks!

Solution

  • I think you definitely have the right thought process and it's hard to say if I should optimize now or later. I would suggest doing it if it's not taking very much time.

    For the case where you have a lot of users, you definitely want your app to track the minimal amount of database fields possible. I think your approach with two subscriptions is good. One thing I would suggest is reducing the amount of subscriptions for the first template by having a publication that has all 10 fields, so it doesn't have to subscribe twice.

    Hope that helps!