Measuring MongoDB I/O and CPU performance

Question

I am new to MongoDB and I'm experiencing this NoSQL db through a series of very simple queries, with an Oracle exported data set.

Aside measuring the execution time with primitives like db.system.profile.find(), I would like to measure I/O performance and system calls. Is there a simple, relevant, way to do this? What would be the best approach?

Also, I have noticed that the size of the collections are way much bigger than in Oracle. For example, an Oracle table (average row length x number of rows) translates to a 5x time bigger collection in MongoDB. Why? Are there any reasons for this?

Thanks for your insights.

Cheers

Answer 1

There are several tools that measure system performance like newrelic,MMS,cacti, ganglia...

Regarding the size you need to share a sample record from Oracle and a sample document from Mongo.

The obvious reasons are:

• long field names: A collection in Mongo with N documents stores the field names N times. If the field names are too long you are wasting space.

• powerof2 allocation strategy (only for MMAP engine): Is now the default record allocation strategy for MMAPv1. With the power of 2 sizes allocation strategy, each record has a size in bytes that is a power of 2 (e.g. 32, 64, 128, 256, 512 ... 2MB). If your document size is 33 then mongo will allocate 64 and you are wasting half your storage.

• Compression (only for MMAP engine): Oracle is using compression while MMAP engine not