Increase max document size to at least 64MB

Initially limit was 4MB, then it was raised to 16MB in 2009.
We are in 2021, we have better hardware, better network, bigger documents and competitors without 16MB limit.
There are a ton of use cases where this limit is too small nowadays, sensors, large documents with history of data.
I’ve opened a new JIRA Issue on this here:


Ticket on Jira has been closed with this comment: “ Thanks for your report. Please note that the SERVER project is for bugs for the MongoDB server. As this ticket appears to be an improvement request, I will now close it.”

But main Jira page for MongoDB says:

Which JIRA project should I use to report bugs or feature requests?

  • To report potential bugs, suggest improvements, or request new features in the MongoDB database server , use Core Server (SERVER) .”

Hi @Ivan_Fioravanti welcome to the community!

Apologies for the confusion. The description of the SERVER project is a bit outdated. We currently are using to collect ideas on how to improve the server, and dedicated the SERVER JIRA project for bug reports. Specifically in your case, you would want to go to the Database Section on that page.

Best regards

1 Like

Hi @Ivan_Fioravanti

Since only data that’s always used together should be stored together, I’m curious about the use case that requires documents bigger than 16MBs. You mention tons of use cases, but can you be a bit more specific? Like “history of data” - this seems like a problematic example since eventually it will outgrow any document size limit, and I’m not sure I can think of a use case where you need all history (no matter how old) when reading a document.


1 Like

Hi @Asya_Kamsky
there are many examples in the
everyone starts using MongoDB thinking: 16MB is a lot! I’ll never hit this limit, but when you reach it is a mess.

Also this one would be extremely beneficial complex aggregations with many pipeline can hit this limit more often than you think.
Removing this limit shoild be easier, please plan at least this one.


That’s because it usually is an indication of incorrect schema design.