Object oriented data modeling - Part 1

Using javascript objects for data modeling.

This article is not about OOP (Object Oriented Programming). This is an article about how I approach modeling databases and api-routes.

Data modeling

What is data modeling?

Data modeling is a crucial practice that involves creating a visual representation of how data is stored, accessed, and managed across different systems. It serves as a blueprint for databases and data management systems, outlining the types of data, the relationships among them, and the rules governing these aspects.

Source: IBM ¹

In this article I will focus on storing, accessing and managing data. The way I visualize data is through javascript objects.

Why use javascript objects for data visualization?

The reason I use a javascript object to visualize dataflow is because it’s easy to understand. It’s also the easiest way to update data when I encounter new requirements. An example of a data object could be an address book.

const basickAddressBook = {
  name: "John Doe",
  web: "jhondoe.com",
  email: "johndoe@jhondoe.com",
  phone: 12345678,
};

Here it’s easy to see and understand what I want to achieve. To save the address book in a database table I will need a column for each key in the object. Because I have added some dummy text I also know what type each column in the database should be.

Name would be of type string. Web would be of type string. Email would be of type string containing a @ and at least 1 dot (.) Phone would be of type numbers. I’m usually not worrying about the database design before I have worked through the whole data workflow. If I start by making the database tables before I have worked through the data requirements I will eventually have to rewrite the table scripts everytime I change the dataobject. That would result in a lot of wasted time and wouldn’t be very productive.

Adding or removing data.

Often, when working on a project, I realize the need to add or remove data from an object. Take the example above. The first version of the address book didn’t include a field for the postal address. If storing an address isn’t necessary, I could stop here and begin designing my database. However, I usually step back to see the bigger picture and find ways to make the system more flexible. By flexible, I mean designing it to accommodate multiple data types and use cases. Later, I can decide which fields should be required and which can be optional. An updated version of the address book might look like this:

const extendedAddressBook = {
  name: "John Doe",
  address: "No Way Street",
  postalCode: "75462",
  city: "Paris, Texas",
  country: "United States",
  web: "jhondoe.com",
  email: "johndoe@jhondoe.com",
  phone: 12345678
};

Here I have added address, postalCode, city and country. The overall object can of course be much more complex. But as far here the address book has become more flexible and can now store postal and internet related information. But there is one thing missing…

What if there are multiple entries for a key value pair?

Let’s say the entry in the address book has more than one phone number or more than one email address. Here I add the emails to an array so it will gives me an array of objects.

const arrayOfObj = [ 
  { email: "email1@example.com" },
  { email: "email2@example.com" }
];

Also I probably have more than one entry in my address book. So the overall object must also be an array of objects. So the current object looks now like the following:

const arrayAddressBookEntries = [{
    name: "John Doe",
    address: "No Way Street",
    postalCode: "75462",
    city: "Paris, Texas",
    country: "United States",
    emailAddresses: [
      { email: "email01@example.com" },
      { email: "email02@example.com" }
    ],
    phonNumbers: [
      { phone: 12345678 }, 
      { phone: 87654321 }
   ]
  }
]

Now I know from the object that the address book can contain multiple entries (the overall array) and each entry can contain multiple emails and multiple phone numbers. If the object had been more complex I would consider splitting up the object into multiple objects so each array is in its own object. When times come to model my database tables I can then use the different objects as templates for the database tables. The three objects looks like this:

const addressBookEntries = [{
    name: "John Doe",
    address: "No Way Street",
    postalCode: "75462",
    city: "Paris, Texas",
    country: "United States"
}];

const emailAddresses = [
  	{ email: "email01@example.com" },
  	{ email: "email02@example.com" }
];

const phonNumbers = [
    { phone: 12345678 }, 
    { phone: 87654321 }
];

Foot notes

www: https://www.ibm.com/think/topics/data-modeling ⤻