How to recycle all server resources in mongodb?

Question

MongoDB

The meaning is as follows: there is a fairly powerful hardware, 128gb, 2 x Xeon E5-2620, 3 TB ride. All this is based on bare Linux and mongodb 2.4.10, the database is large, about 320GB, 100 million records. Only writes are performed and mapreduce processes are constantly running. It is clear that mr processes do not use all 2 x 2 x 4 cores .. Therefore, the question is how to dispose of all this stuff so that the processor does not stand idle but is engaged in useful activities. Something in my head except how to run 16 virtualboxes and make a replica out of them, nothing comes up.
Maybe there are more humane ways to use all the cores? Or everyone does.
PS: aggregation framework that can eat kernels is not suitable at all. Need MR tasks.

Answer 1

Well, let's start with the fact that the option with replicas and virtual machines is outright nonsense and a waste of resources, it would be okay to shard, then it would be possible to parallelize map-reduce (in theory) with the right approach. Yes, and virtual machines are not needed, you can simply raise several instances on different ports.
In general, the topic is quite hackneyed. In particular, there is nothing corny to load the CPU with, everything rests on IO.
https://groups.google.com/forum/#!topic/mongodb-us...

Answer 2

If you have a lot of parallel requests, then it uses all the resources itself, do you have the database on ssd disks?