Is there much of a performance boost from using observeChanges over observe for a Mongo cursor? In particular, does observe benefit from tailing Mongo’s oplog?
The oplog announcement (https://github.com/meteor/meteor/wiki/Oplog-Observe-Driver) only mentions observeChanges, so does that mean observe continues to rely on polling?
Thanks in advance!
Did you find out? I’m interested in this question as well. If I understand it right this is especially true if you have a large document (ever a lot of fields or some large content in a field) and you change just one tiny flag from true to false then observe will send the whole document over the network.
On the contrary if the field that is changed is the only field in the document, there should be no difference between the amount of data that is transferred via observe or observeChanges.
Is that also your understanding?