What is the relationship between map-reduce and nosql?

Question

I want to use nosql for my application. The purpose of nosql is to store user log data, to use and analyze the data, and to provide customized data to users. Here we came to know about map-reduce in search of algorithm and method to process large amount of log data quickly.

I have a few questions:

1. Is map-reduce an algorithm?

2. Is map-reduce suitable for fast processing of large amounts of data?

3. How can I use nosql in addition to map-reduce for faster speed?

4. I know that mongodb supports map-reduce, is that correct?

I do not understand exactly the relationship between nosql and map-reduce.

Thanks.

Answer 1

1. NoSQL = [Not only SQL] database are the types of databases that can have structural, semi-structural(XML, json) or non-structural data(textual data).

2. Yes it can help for processing large data sets.

3. Where as Map-reduce is an algorithm. Please read this article to understand how map reduce works in NoSQL or big data applications.

EDIT

Here is some good resource for learning mapReduce & Big Data technologies. BTW these tutorials are in hindi.

Answer 2

Is map-reduce an algorithm?

MapReduce is not exactly an algorithm, rather a tool which can be used with many algorithm, which make a good "fit". Mapreduce leverage the features of hadoop distributed data storage and processing. As you may have notice, not all the algorithm can be "efficiently" implemented using mapreduce . So, a design decision should be made based on various factors like data volume, processing restrictions etc.

Is map-reduce suitable for fast processing of large amounts of data?

mapreduce does a lot of disk I/O during its processing and hence is not suitable for the cases, where execution time is a constraint. You may want to switch to spark for faster processing. Using tez engine with mapreduce is another option. However, do not compare mapreduce performance with nosql database like hbase . mapreduce and nosql both belong to two entirely different technology stack.

How can I use nosql in addition to map-reduce for faster speed?

It depends on your use case. It is very common to process hbase data into a mapreduce program to produce analytical results.

I know that mongodb supports map-reduce, is that correct?

Let me re-phrase it. mapreduce is a tool for which mongodb may be a data source.