As suggested, I’m making a separate post about this. I wasn’t going to initially since I thought it might just be me, but it seems like there are similar things going on with other people.
I’ve been developing an app using MongoDB Realm with Device Sync. It’s still in development with only a few users testing, that too only periodically. The only connection to my database is through Device Sync. My Atlas and database instance are on Google Cloud us-central1. The total size of the data in my database is only 711 KB. Yet somehow, I have monthly data transfer rates of 20 GB and 80 GB. How is this at all possible when there are essentially no active users and the only purpose is for development and testing?
Additionally, in Atlas Metrics it says only 84,483.1 B of data has been transferred for the month. How is this number in any way extrapolated to 20 GB and 80 GB?
I also noticed in the Metrics and Real Time view there are constant operations, bytes in and out, 40-60 constant connections when no one is using the app, and in some cases, queries constantly being run.
They don’t align with Device Sync logs. I was unable to find an answer so I’m on the verge of abandoning Atlas and Device Sync altogether and just using a local Realm, but just in case there is a solution, making this post.
Just wanted to double check, is it possible this usage is standard for a cluster and not expected to continue to scale exponentially with additional usage/users?
I was wondering if this usage and constant communication is just related to the database internals or shard communication and is more of a “flat” fee that is incurred and required and not necessarily related to usage from Sync or incoming reads/writes.
I hope that makes sense. If not, I’ll try to explain better.
Hi Akansh,
Thanks for giving some time to look at this.
You’re correct that this is a just a “flat” baseline usage of the cluster and is just a result of internal transfer activity. When device sync is enabled on atlas, it does require ongoing network transfer in the background to coordinate real-time updates with the cluster, even when usage is mostly idle. From a billing standpoint, this transfer cost is nearly negligible (generally less than a dollar a month).
Data transfer will increase as the app activity increases, but that volume will primarily be proportional to the amount of data that is queried and fetched by connected clients, which depends entirely on the workload of the app itself - and independent of the baseline transfer activity for the app.
