Hello

Use the following criteria:

Overhead

Response Time

Platform Support

Programming Model

Performance

Ease of Use

Compatibility

Data Processing

Failure Tolerance

Security

More is explained in the links below.

https://www.xplenty.com/blog/2014/11/apache-spark-vs-hadoop-mapreduce/

http://www.infoworld.com/article/3014440/big-data/five-things-you-need-to-know-about-hadoop-v-apache-spark.html

Hello

Use the following criteria:

Overhead

Response Time

Platform Support

Programming Model

Performance

Ease of Use

Compatibility

Data Processing

Failure Tolerance

Security

More is explained in the links below.

https://www.xplenty.com/blog/2014/11/apache-spark-vs-hadoop-mapreduce/

http://www.infoworld.com/article/3014440/big-data/five-things-you-need-to-know-about-hadoop-v-apache-spark.html

You can use these links :

http://www.naturalspublishing.com/files/published/v6910010rnl56m.pdf

https://hal.inria.fr/hal-01199200/document

http://www.kdnuggets.com/2016/03/top-big-data-processing-frameworks.html

You can use the parallel Scaleup and Speedup  measures. These measures are used to evaluate the performance of scalability of the algorithms. 

Check my publication:

https://www.researchgate.net/publication/233910085_Parallel_Particle_Swarm_Optimization_Clustering_Algorithm_based_on_MapReduce_Methodology

Conference Paper Parallel Particle Swarm Optimization Clustering Algorithm ba...