Kadira response time (Mongolabs free vs. $15 plan)



I’m checking out MongoLab’s free plan (no oplog), and I notice there’s about a 400-500ms response time lag in Kadira. Is this normal whenever the database is hosted by another service? And does anyone have any experience with the $15 service also from MongoLab (1GB to 8GB, with oplog)?


You need to look at some traces to see what’s the exact issue here on the latency.

When you select a DB you also need to choose it near to your app server as well. All these mongodb players have a solution to choose the data center. Try that too.


@arunoda ah ok. Let me see here. Well, the server I got is based in Italy, so Mongolab’s 3rd party supplier options include only Ireland and I Netherlands (some place in western Europe or something else). I tried the one based in Ireland thinking it’d be in the same area and one of the closer ones to the server.

Now for the traces, I see that for example, the login method shows me that DB’s response time is usually about 400-600ms, sometimes going over 800. There’s one point where it went up to 6s lol. Now when clicking on something that has a 800+ms response time for the DB, I see in traces that:

  • There are some observeChanges that spike to about 140ms. Most of the time they’re around 40ms. Since I’m using the throttle package, there’s a bunch of those taking up 40ms each.

There’s no users at the moment, but should I be worried about these numbers, or you think that even with 5 to a couple of hundred users, there’s a possibility it won’t shoot up immediately and the system will be able to serve those users albeit with this kind of base performance?


I don’t you don’t need to worry right now. Those figures are pretty decent. But, it’s great if you can get the servers into the same datacenter. (Not the same country/region)


I suppose I’ll have to move the database locally later on if it’ll slow down. I don’t think there’s third party option unless I go for rackspace, which is kinda pricey atm. My question is, how do I enable the oplog if I’m going to run only 1 instance of the mongo database? Or can I use something like a local database + one or two more of these MongoDB service providers as part of the oplog redundant “chain”?