监控服务MongoDB架构设计

Question

I have 1000+ streaming audio devices deployed around the world. All of which currently check in with the factory monitoring service about every 15 minutes. I'm rolling my own monitoring service and will be storing the status updates with mongo.

Question:

What would be a better design schema:

1) One massive collection named device_updates . Where all the status update document would include a device serial_number key?

2) 1000+ collections each named with the devices serial number, ie: 65FE9 , with the devices status updates siloed in their own collection If going this route I would cap the collections at about 2000 status update documents.

Both would need to be indexed by the created_at date key.

Any ideas on which would be better performance wise? Or any thoughts on what would be the preferred method?

Thanks!

Answer 1

I would definitely go for one massive collection , since all the documents are of the same type .

As a general rule, think of a collection in MongoDB as a set of homogeneous documents . Having just one collection, moreover, makes it much easier to scale out horizontally (ie, by use of shards ), by using for example the serial_number as the shard key.