3

I am new with MongoDB sharding, and we are facing performance issues. I tried with many test cases with three servers and I feel MongoDB sharding does not give better performance compare to single instance of mongod on a single server.

Am I correct? If I am not correct then kindly provide test collection, shard key, and load test procedure.

Test case 1:

I did one load test between single instance on a mongod and single shard (http://docs.mongodb.org/manual/core/sharded-cluster-architectures-test/) and I observed 40% performance degradation in sharding setup because of mongos and config server.

Does mongos and config server take 40% server resources?

Test case 2:

We observed 10K TPS on single instance of mongod instance. We observed 5.1K TPS in mongo shard on three servers with same capacity hardware (we used same server for single mongod instance test). We used shard key as hashed index ObjectId.

Why we could not get 10K+ performance with three servers? Is there any way to get 10K + performance using sharding? If yes then how?

Improve this question
2
  • 1
    Foursquares uses Mongo to store millions of daily checkins. So it must be possible. Maybe this presentation would give you some insight, which you could use to make your question more precise (less broad). Commented Jul 15, 2013 at 13:41
  • 2
    it's is not the readers' job to provide you test collection, shard key and load procedure. This is not really an appropriate forum of this type of question - it's not about programming and it doesn't have a simple answer. Commented Jul 17, 2013 at 5:25

1 Answer 1

Reset to default
-10

Use couchbase instead. We stopped using MongoDB due to it's poor performance compared to what is out there.

Improve this answer
Sign up to request clarification or add additional context in comments.

7 Comments

LOL! Thats all I gotta say.
Hmm, any evidence it wasn't your own design faults that caused this? You provide no real substantial evidence yet we are to believe you that MongoDB is just slow and poor performance? If I am honest this is not an answer, don't know who upvoted...
It's easy to find evidence to show the performance issues. Here is one example:infoq.com/news/2013/04/NoSQL-Benchmark
As stated concerns in the comments there is some problems with that benchmark, mainly some information missing on its details. I do hate benchmarks because they rarely refelct what YOU will see, and who is to say they used the tech correctly too. Anyway I am fairly (99%) certain I get better figures on MongoDB than they did
There was no question that migrating over to couchbase from MongoDB increased performance when scaling enormously. You should try it, you won't look back. Pencils have erasers for a reason, people make mistakes.
|

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Start asking to get answers

Find the answer to your question by asking.

Ask question

Explore related questions

See similar questions with these tags.