AWS DocumentDB with Parse Platform

Dear Parse Community,

For one of our projects we are currently considering a noSQL solution using Parse platform on top of AWS DocumentDB. There is not much information available on the web with regards to Parse coupled with DocumentDB.

In theory, it would suffice to “just point Parse to DocumentDB” as the latter is “Mongo 3.6 compatible”. I would not expect it to be this smooth in real life. My question is: Does anyone have (real life or test) experience with this combination?

I am particularly interested in the following two questions:

  • Are there any compatibility limitations in AWS DocumentDB versus MongoDB that affect Parse?
  • Is there a performance impact for using Parse with DocumentDB vs Parse with Mongo?

I am happy to share an update once I have tested the combination myself but would be grateful to hear from those who have tried this already. Alternatively, MongoDB Atlas is an option but I am particularly interested in DocumentDB in this case.

Thank you!

The feature set of AWS DocumentDB is different from that of MongoDB, for example geospatial queries are not supported at all.

It really depends on your use case whether DocumentDB is a technical fit for you. It’s tempting to use DocumentDB for its abstraction and AWS integration but you could be up for a surprise when you suddenly find a feature that is not supported. A safe way would be to map out the features you expect from your DB now (and possibly in the near future) and look into the DocumentDB docs which offers a convenient table for feature comparison between MongoDB and DocumentDB and maybe contact the AWS DocumentDB support team for a detailed inquiry.

I assume you could search online for general price / performance comparisons between MongoDB and DocumentDB. It could depend on whether your app is more read or write heavy as AWS DocumentDB has some conceptual optimizations for writes.

At this point, I believe that neither the estimated price difference nor DocumentDB’s write-ahead log feature would make much difference in our case. Thank you for pointing out the optimisation though, it could be worth considering in other use cases.

That’s definitely a helpful overview. The lack of geospatial queries in DocumentDB is indeed a deal-breaker at this point.

Thank you for your input!