Home/ Questions/Q 8021703
In Process

The Archive Base Latest Questions

Editorial Team
  • 0
Asked: 2026-06-04T22:03:50+00:00

I’ve to perform a map reduce operation with a clustering algorithm of a large

  • 0

I’ve to perform a map reduce operation with a clustering algorithm of a large amount of data.
I’ve choose MongoDB for its scalability, great docs, BSon documents storage and many other great features.

But, the map-reduce operation is too slow. Only one core of my quadcore CPU is used.

CPU utilization
Disk Utilization

I read this document and I understand that is a JavaScript’s fault.

“Only one thread in the mongod process executes Javascript at a time”

How can I improve that? Improving IO speed?

  • SSD storage may be useful?
  • Load the whole MongoDB’s data on RAM?

Otherwise bottleneck is only the CPU and the mono-thread algorithm, so I have to re-write my algorithm in another language, outside Mongo, to take advantage of multi-threading and multi-processing?

Share

Leave an answer

You must login to add an answer.

Need An Account,

1 Answer

  1. Editorial Team

    Otherwise bottleneck is only the CPU and the mono-thread algorithm

    The issue is that there is no parallelization of work while Mongo is using Spidermonkey as its js engine. There is a long term plan to move to V8 to allow multi-threaded execution of js tasks, but it won’t make the next release (2.2).

    Some of the slowness with map/reduce will be improved in 2.2 for people who will be able to switch to the new aggregation framework instead of map/reduce. Unfortunately it doesn’t sound like that will help you since you are using custom algorithm and not just aggregation computations.

    • 0