In this post we can create a Web API sample using Cosmos DB.

(this is similar to the MVC Sample of MS)

Step 1: Create ASP.NET Core web application

Step 2: Choose API

Step 3: Add Controller

Name it as ItemController

Step 4: Add Nuget Package

Step 5: Create Class to represent Item

namespace CosmosDB_WebAPI.Models
{
     public class Item
     {
         [JsonProperty(PropertyName = “id”)]
         public string Id { get; set; }

        [JsonProperty(PropertyName = “name”)]
         public string Name { get; set; }

        [JsonProperty(PropertyName = “description”)]
         public string Description { get; set; }

        [JsonProperty(PropertyName = “isComplete”)]
         public bool Completed { get; set; }
     }
}

Step 6: Modify Controller Code

Add the following content to the ItemConroller to include GET, POST, PUT, DELETE functionalities.

namespace CosmosDB_WebAPI.Controllers
{
     [Route(“api/[controller]”)]
     [ApiController]
     public class ItemController : ControllerBase
     {
         private readonly ICosmosDbService _cosmosDbService;
         public ItemController(ICosmosDbService cosmosDbService)
         {
             _cosmosDbService = cosmosDbService;
         }

        // GET api/values
         [HttpGet]
         public async Task<IEnumerable<Item>> Get()
         {
             return await _cosmosDbService.GetItemsAsync(“SELECT * FROM c”);
         }

        // POST api/values
         [HttpPost]
         public async void Post([FromBody] Item item)
         {
             item.Id = Guid.NewGuid().ToString();
             await _cosmosDbService.AddItemAsync(item);
         }

        // PUT api/values/5
         [HttpPut(“{id}”)]
         public async void Put(string id, [FromBody] Item item)
         {
             await _cosmosDbService.UpdateItemAsync(id, item);
         }

        // DELETE api/values/5
         [HttpDelete(“{id}”)]
         public async void Delete(string id)
         {
             await _cosmosDbService.DeleteItemAsync(id);
         }
     }
}
Step 7: Modify Config class

Configuration file below.

{
   “AllowedHosts”: “*”,
   “CosmosDb”: {
     “Account”: “https://localhost:8081″,
     “Key”: “C2y6yDjf5/R+ob0N8A7Cgv30VRDJIWEHLM+4QDU5DE2nQ9nDuVTqobD4b8mGGyPMbIZnqyMsEcaGQy67XIw/Jw==”,
     “DatabaseName”: “Tasks”,
     “ContainerName”: “Items”
   }
}

Step 8: Run the application

Step 9: Test with Postman

GET https://localhost:44364/api/item

POST https://localhost:44364/api/item

{
     “name”: “Item 2”,
     “description”: “Description 2”,
     “isComplete”: true
}

PUT https://localhost:44364/api/item/1

{
     “id”: “1”,
     “name”: “Item (modified)”,
     “description”: “Description (modified)”,
     “isComplete”: true
}

DELETE https://localhost:44364/api/item/566b8151-3fc4-4138-8a2f-22b6939ee60a

Download

You can download the sample here.

As you know Azure Cosmos DB will cost high if you are spending time on learning & experimenting.  Alternatively, following are different options inside/outside the Azure Portal to try with:

MSDN $200 Free Plan

https://azure.microsoft.com/en-us/free/free-account-faq/

Emulator

https://aka.ms/cosmosdb-emulator

Sandbox

https://docs.microsoft.com/en-us/learn/modules/create-cosmos-db-for-scale/2-create-an-account

Cosmos DB offer multiple models as below:

• Document Model with SQL API
• MongoDB Model with MongoDB API
• Cassandra Model with Cassandra API
• Table Model with Table API
• Graph Model with Gremlin API

In this article we can see which model to choose based on the requirement.

Note As informed previously we are making a Choice of CosmosDB here due to Dynamic Schema requirement & Geo-replication.

Document Model with SQL API

If you are starting with a new project which requires storage of similar items, schema changes a lot, then SQL API is the best choice.

MongoDB with MongoDB API

If you are Migrating from existing MongoDB which have Investments of MongoDB Queries then MongoDB API is the best choice.

Cassandra Model with Cassandra API

If your project requires Web Analytics capabilities, then Cassandra API is the right choice.

Table Model with Table API

If your data consists of Key-Value pairs OR if you wanted to Migrate data from Azure Table Storage, then Table Model API would be right choice.

• Cosmos DB (access based pricing) is Cheaper than Table Storage (size based pricing) depending on the case

Graph Model with Gremlin API

If you have relationships between data, then Graph Model is the best choice.

Example: Amazon wanted to show Product Recommendations like customer who bought LG TV 46” also bought another product TV Stand 46”

graph.V().hasLabel(‘product’).has(‘productName’, ‘LG TV’).addE(‘boughtWith’).to(g.V().hasLabel(‘product’).has(‘productName’, ‘TV Stand’))

As discussed earlier, the Decision for choosing CosmosDB rather than SQL Server is mostly one of the following:

• Schemaless Advantage
• Geo Replication

In this post we can explore the same with few scenarios.

Player Table

Let us think of a Table named Player in SQL Server.  There will be properties like following:

• ID
• Name
• Age
• Address

Requirement 1

Now, the Player table NEED TO support both Football player & Cricket player.

We resolve this by adding a Type column to differentiate between the players.

Requirement 2

The requirement changed that:

• The Football Player should have a Goals column of type int
• The Cricket Player should have a Run Rate column of type float

We resole this by adding 2 more columns. 

Problems in Design

As you can see, there are few problems in current design:

• There is unwanted column GOALS being allocated for CRICKET Player
• There is unwanted column RUN RATE being allocated for FOOTBALL Player

The problem is more complicated when the FUTURE REQUIREMENT to support Tennis Player, Badminton Player also comes through.

How COSMOS DB resolves this issue?

Cosmos DB resolves this issue by using only required columns.

{

“name”: “John”,

“age”: “30”,

“type”: “Football”,

“goals”: 4,

}

{

“name”: “Kevin”,

“age”: “25”,

“type”: “Cricket”,

“runrate”: 5.3

}

Summary

In this post we have observed the Schemaless advantage of Cosmos DB.