M2Bench: A Database Benchmark for Multi-Model Analytic Workloads

• M2Bench is a new benchmark program capable of evaluating a multi-model DBMS.
• M2Bench supports several important data models including relational, document-oriented, property graph and array models.
• M2Bench consists of multi-model workloads that are inspired by real-world problems.
• Each task of the workloads mimics a real-world problem that involves at least two different models of data.

Datasets

• All datasets in M2Bench are explained in the Datasets folder.

Workloads (Tasks)

• All tasks in M2Bench are explained in the Tasks folder.

How to Run M2Bench

• Generate datasets of M2Bench following the instructions in the Datasets folder.
• Choose DBMS X [agensgraph|arangodb|orientdb|polyglot].
• Load the datasets into DBMS X following the instructions in the Impl/X/load_datasets folder.
• Run tasks of M2Bench following the instruction in the Imp/X/run_tasks folder.

Directory Overview

m2bench
├── Datasets                #How to generate the datasets of M2Bench from the realworld datasets.
│   ├── ecommerce          
│   ├── healthcare
│   ├── disaster
│   └── README.md
├── Tasks                   #Description of all tasks of M2Bench
|   └── alltasks.md
├── Impl                    #Implementation of tasks of M2Bench on various DBMSs
│   ├── agensgraph  
│   │   ├── load_datasets   #How to load datasets into DBMS
│   │   └── run_tasks       #How to run tasks
│   ├── arangodb
│   │   ├── load_datasets 
│   │   └── run_tasks
│   ├── orientdb
│   │   ├── load_datasets 
│   │   └── run_tasks
│   └── polyglot            #polyglot persistence
│       ├── load_datasets
│       └── run_tasks
├── README.md
└── LICENSE