I recently implemented Oplog tailing in my Meteor.js app using my MongoLab cluster to enhance performance, availability, and redundancy. However, I have noticed that since incorporating this feature, my publications are taking longer to complete. While a delay of around 200ms may not seem significant, in many cases it is taking much longer. This is particularly evident in a specific publication I have discussed here.
The response time for this publication is already lengthy, and the additional delay caused by oplog observations is further impacting it. This issue is not isolated to only one publication, as I have observed similar delays in other areas where oplog observations are involved.
I have searched extensively online for an explanation as to why observing oplogs is slowing down my publications, but have been unable to find any information. If anyone can shed light on this issue, I would greatly appreciate it.
Below are some Kadira screenshots that visually represent the performance disparities:
https://i.sstatic.net/UW5Zp.png
Here is a snapshot from another pub/sub:
https://i.sstatic.net/Lfeir.png
Lastly, an example where oplog observations have a reasonable impact on performance (though still affecting the pub/sub):