Comparison of IRIS Native Globals API and EAV-approach in relational db (MySQL)

About EAV-model

The EAV-model is designed to store hierarchical and sparse structures in relational databases.

The EAV-approach is widespread among programmers, but with the advent of IRIS Native API, is it worth using it further?

My articles about EAV-model and IRIS globals

Entity-attribute-value model in relational databases. Should globals be emulated on tables? Part 1.

Entity-attribute-value model in relational databases. Should globals be emulated on tables? Part 2.

Wikipedia

About EAV-model

Purpose of this project

Make a comparison of the performance approaches on the numbers. I want to see how much time it takes to create thousands of products, as well as to access them using IRIS Native API and EAV (MySQL).

In the project, a data scheme is created the same as in real projects on IRIS globals and SQL for objective comparison.

Transactions are used in both approaches. Products are created in random sections of the catalog.

When testing the loading of product properties, random products are loaded to make it harder to use caching database techniques.

Data model in project

The structure of the demo data that we will store:

Global for this structure created in beginning schema.py.

SQL schema schema.sql.

Original tools inside my project

I created a usefull library in Python iglobal.py that allows you to work with globals in an object-oriented way.

node = good.node('index1', 'index2')
node.set(5)

node.inc(2)

i = node.iterator()
print(node.get())
good.kill()

You may initialize whole tree with one operation:

good = iGlobal('good')
good.init(

{'Storages':

{ 'id' : 1,

'Properties':

{'capacity':

{'name': 'Capacity, GB',

'sort': 100,

'searchable': 1,

'table_view': 1

},

'weight':

{'name': 'Weight, kg',

'sort': 800,

'searchable': 1,

'table_view': 1

}

},

'SSD':

{'id' : 2,

'Properties':

{'endurance':

{'name': 'Endurance, TBW',

'sort': 200,

'searchable': 1,

'table_view': 1

}

},

'AIC PCI-E': {'id': 3},

'SATA':      {'id': 4},

'M.2':       {'id': 5}

}

}

}

)

Estimated container build time

First time running ~2-15 minutes. Depending on the speed of your internet. After the 1st time running the next ones will perform better and take few seconds.

Usage

Build containers

docker-compose build

Run containers

docker-compose up -d

Run application in browser

Go to localhost:5000

Run application in command line

After starting containers open another terminal (under root user) and type commands:

docker ps

See IRIS container name (column NAMES)

# Go in IRIS container
docker exec -it IRIS_CONTAINER_NAME bash
# inside container
cd /app
python3 cmd.py

Stop containers

docker-compose stop