Skip to main content

Implementing DataLoader For Query Execution Using Hot Chocolate GraphQL In .Net

In this article, we are going to understand and implement the DataLoader concept in Hot Chocolate GraphQL.

Here we will use an existing Hot Chocolate sample application that was created in the introduction articles on Hot chocolate GraphQL. I recommend reading those articles:
Part-1 Introduction On Hot Chocolate GraphQL

Issues Need To Fix:

Before implementing the DataLoader into our sample we must aware of the issues we are facing. In GrphQL we have an option like Fragments Query. Fragments query is used to display the comparison results very effective way.
Fragments Query:
query{
  Gadget1:GetByBrand(brand:"samsung"){
    Id
    ProductName
    Brand
  }
   Gadget2:GetByBrand(brand:"red mi"){
    Id
    ProductName
    Brand
  }
}
Here we request a GraphQL endpoint to serve the different gadgets for comparison this type of query called Fragments Query.

Now this kind of Fragments Query produces issues like:
  • We will face a threading issue with the database context on using the async database communication calls.
  • In a normal process, this Fragments Query will run the query resolver method(logic that contains database communication) count is like based on n-number of comparison queries. This means if you observe the above sample query we requesting 2 comparison results so the revolver method will execute 2 times(which means 2 database calls). This might leads to performance issues.

Async Calls Database Context Issue:

Let's expose ourselves to the kind of issue we will get using the asynchronous calls.

In Part-2  written a resolver method as synchronous like below.
SchemaCore/Query.cs:(Existing Logic)
public List<Gadgets> GetByBrand(string brand, [Service] MyWorldDbContext context)
{
	return context.Gadgets.Where(_ => _.Brand.ToLower() == brand.ToLower()).ToList();
}
The results look as below.

Now let's update our resolver to an asynchronous approach.
SchemaCore/Query.cs:(Existing Logic)
public async Task<List<Gadgets>> GetByBrand(string brand, [Service] MyWorldDbContext context)
{
  return await context.Gadgets
  .Where(_ => _.Brand.ToLower() == brand.ToLower()).ToListAsync();
}
Let's check the results for the asynchronous calls.
You can see we are getting errors in the result set. But the actual error is like this:-

An exception occurred while iterating over the results of a query for contexttype'HC.GraphQL.Sample.Data.Context.MyWorldDbContext'.
System.InvalidOperationException: A second operation was started in this context before a previous operation completed. This is usually caused by different threads concurrently using the same instance of DbContext. For more information on how to avoid threading issues with DbContext.

This error because in the fragment query we are requested 2 comparison results, since our resolver is an asynchronous resolver so 2 requests are executed parallel and both are trying to access the database context at the same time.

Use Database Context Pool To Resolve The Threading Issue:

In .net DbContext pool keeps by default 128 Dbcontext alive in the pool. So that these instances will be served to requests without destroying them immediately. So if our fragment request will use separate DBContext for each thread request.

Existing DbContext registration looks like as below:
Startup.cs:(Existing Code)
services.AddDbContext<MyWorldDbContext>(options => {
  options.UseSqlServer(Configuration.GetConnectionString("MyWorldDbConnection"));
});
Now let's update our code to register the DbContext pool as below:
Startup.cs:(Updated Code)
services.AddPooledDbContextFactory<MyWorldDbContext>(options => {
  options.UseSqlServer(Configuration.GetConnectionString("MyWorldDbConnection"));
});
Now let's implement our logic to fetch the DbContext instance from the pool of DbContext. So let's create an extension method of the IObjectFieldDescriptor that will generate DbContext as below.
Extensions/ObjectFieldDescriptorExtensions:
using HotChocolate.Types;
using Microsoft.EntityFrameworkCore;
using Microsoft.Extensions.DependencyInjection;

namespace HC.GraphQL.Sample.Extensions
{
  public static class ObjectFieldDescriptorExtensions
  {
   public static IObjectFieldDescriptor UseDbContext<TDbContext>(
	this IObjectFieldDescriptor descriptor
   ) where TDbContext: DbContext
   {
	return descriptor.UseScopedService<TDbContext>(
		create: s => s.GetRequiredService<IDbContextFactory<TDbContext>>().CreateDbContext(),
		disposeAsync: (s,c) => c.DisposeAsync()
	);
   }
  }
}
  • (Line: 14) Fetching the context from the pool.
  • (Line: 15) Releases the allocated resource to DbContext.
Now let's create an Attribute which we are going to decorate on resolver so that they receive scoped DbContext from the pool.
Attributes/UseMyWordDbContextAttribute.cs:
using System.Reflection;
using HC.GraphQL.Sample.Data.Context;
using HotChocolate.Types;
using HotChocolate.Types.Descriptors;

namespace HC.GraphQL.Sample.Extensions
{
  public class UseMyWorldDbContextAttribute : ObjectFieldDescriptorAttribute
  {
   public override void OnConfigure(IDescriptorContext context, 
   IObjectFieldDescriptor descriptor, MemberInfo member)
   {
		descriptor.UseDbContext<MyWorldDbContext>();
   }
  }
}
  • (Line: 8) Inherits from the 'HotChocolate.Types.ObjectFieldDescriptorAttribute'.
  • (Line: 13) Above create generic extension method registered here with our database context.
Now update our resolver method to decorate the attribute on it as below.
Schemas/Query.cs:
[UseMyWorldDbContext]
public List<Gadgets> GetByBrand(string brand, [ScopedService] MyWorldDbContext context)
{
  return context.Gadgets.Where(_ => _.Brand.ToLower() == brand.ToLower()).ToList();
}
  • (Line: 1)Decorated our 'UseMyWorldDbContext' attribute
  • (Line: 2) Used the 'ScopedService' attribute to inject the Dbcontext into our resolver method.
Now let's run our fragments query on our asynchronous resolver which will now work perfectly without any threading issue.

Fragment Query Multiple Database Calls Issue:

As mentioned that fragment query will run the resolver method multiple times based on the number of comparison queries inside of it. For the above sample query or screenshots, we used only 2 comparison queries inside of the fragment query so this will invoke the resolver method 2 times which might be performance issues. 

From the above 2 images, we can observe resolver executed 2 times.

DataLoader To Fix Multiple Database Calls Issue:

DataLoader technique in GraphQL to fetch the fragments of data with a single execution of the resolver method which gives us performance benifits.

Let's implement a DataLoader for our sample.
DataLoader/GadgetsByBrandDataLoader.cs:
using System.Collections.Generic;
using System.Linq;
using System.Threading;
using System.Threading.Tasks;
using HC.GraphQL.Sample.Data.Context;
using HC.GraphQL.Sample.Data.Entities;
using HotChocolate.DataLoader;
using HotChocolate.Fetching;
using Microsoft.EntityFrameworkCore;

namespace HC.GraphQL.Sample.DataLoader
{
  public class GadgetsByBrandDataLoader : BatchDataLoader<string, List<Gadgets>>
  {
	private readonly IDbContextFactory<MyWorldDbContext> _dbContextFactory;
	public GadgetsByBrandDataLoader(IDbContextFactory<MyWorldDbContext> dbContextFactory,
	BatchScheduler batchScheduler) : base(batchScheduler)
	{
	  _dbContextFactory = dbContextFactory;
	}
	protected override async Task<IReadOnlyDictionary<string, List<Gadgets>>> LoadBatchAsync(IReadOnlyList<string> keys, CancellationToken cancellationToken)
	{
	 await using (MyWorldDbContext dbContext = _dbContextFactory.CreateDbContext())
	 {
	  var gadgets = await dbContext.Gadgets
	  .Where(g => keys.Select(_ => _.ToLower()).ToList().Contains(g.Brand)).ToListAsync(cancellationToken);
	  var result = gadgets.GroupBy( _ => _.Brand.ToLower())
	 .Select(_ =>new {
		key = _.Key.ToLower(),
		gadgets = _.ToList()
	 }).ToDictionary(_ => _.key, _ => _.gadgets);
	  return result;
	 };
	}
  }
}
  • (Line: 13) To create a HotChocolate GraphQL Dataloader we need to inherit base class DataLoader that is 'HotChocolate.DataLoader.BatchDataLoader<TKey,TValue>'. Here 'TKey' is the type of our query parameter value and 'TValue' is the output type returned from the delivery.
  • (Line: 16&17) Injected 'HotChocolate.Fetching.BatchScheduler' and then its value passed to the base constructor. Also injected 'Microsoft.EntityFrameworkCore.IDbContextFactory'.
  • (Line: 21) Implemented the 'LoadBatchAsync' method whose return type is key-value pair dictionary.
  • This method has input parameters is like collection string as the variable name 'keys', all these collections of string nothing but our query parameters from each comparison query this helps to fetch the data from the database with a single attempt.
  • (Line: 25-31) Fetching results based on the input parameter collection 'keys' and return the result as dictionary data.
Now register our DataLoader in Startup.cs file.
Startup.cs:
services.AddGraphQLServer()
.AddQueryType<QueryObjectType>()
.AddMutationType<MutationObjectType>()
.AddDataLoader<GadgetsByBrandDataLoader>();
Now let's implement our resolver method where we are going to use our DataLoader to output the results.
SchemaCore/Query.cs:
public async Task<List<Gadgets>> GetByBrandLoader(
  string brand,
  GadgetsByBrandDataLoader dataLoader,
  CancellationToken cancellationToken)
{
   var result = await dataLoader.LoadAsync(brand,cancellationToken);
   return result;
}
  • Here for our resolver first parameter is our query parameter and the second is our Dataloader passed as a parameter this will be passed by the framework automatically.
Now let's create a GraphQL field for our new resolver method.
ObjectTypes/QueryObjectType.cs:
descriptor.Field(g => g.GetByBrandLoader(default, default, default)).Type<ListType<GadgetsObjectType>>()
Name("GetByBrandLoader");
Now let's test our fragments query that will return the result with a single time execution of the resolver method with the help of DataLoader.
That's all about the DataLoader in the HotChocolate GraphQL.

Support Me!
Buy Me A Coffee PayPal Me

Wrapping Up:

Hopefully, I think this article delivered some useful information about DataLoader in GraphQL. I love to have your feedback, suggestions, and better techniques in the comment section below.

Refer:

Follow Me:

Comments

  1. Nice article, but I don't see the difference between the existing and the updated DbContext registration. Am I missing something?

    ReplyDelete
  2. The updated DbContext registration should be this (I think, the author copied the code but forgot to change the method name):

    services.AddPooledDbContextFactory(options => {
    options.UseSqlServer(Configuration.GetConnectionString("MyWorldDbConnection"));
    });

    The article itself is awesome, thanks for it.

    ReplyDelete

Post a Comment

Popular posts from this blog

Blazor WebAssembly Custom Authentication From Scratch

In this article, we are going to explore and implement custom authentication from the scratch. In this sample, we will use JWT authentication for user authentication. Main Building Blocks Of Blazor WebAssembly Authentication: The core concepts of blazor webassembly authentication are: AuthenticationStateProvider Service AuthorizeView Component Task<AuthenticationState> Cascading Property CascadingAuthenticationState Component AuthorizeRouteView Component AuthenticationStateProvider Service - this provider holds the authentication information about the login user. The 'GetAuthenticationStateAsync()' method in the Authentication state provider returns user AuthenticationState. The 'NotifyAuthenticationStateChaged()' to notify the latest user information within the components which using this AuthenticationStateProvider. AuthorizeView Component - displays different content depending on the user authorization state. This component uses the AuthenticationStateProvider

How Response Caching Works In Asp.Net Core

What Is Response Caching?: Response Caching means storing of response output and using stored response until it's under it's the expiration time. Response Caching approach cuts down some requests to the server and also reduces some workload on the server. Response Caching Headers: Response Caching carried out by the few Http based headers information between client and server. Main Response Caching Headers are like below Cache-Control Pragma Vary Cache-Control Header: Cache-Control header is the main header type for the response caching. Cache-Control will be decorated with the following directives. public - this directive indicates any cache may store the response. private - this directive allows to store response with respect to a single user and can't be stored with shared cache stores. max-age - this directive represents a time to hold a response in the cache. no-cache - this directive represents no storing of response and always fetch the fr

.Net5 Web API Managing Files Using Azure Blob Storage

In this article, we are going to understand the different file operations like uploading, reading, downloading, and deleting in .Net5 Web API application using Azure Blob Storage. Azure Blob Storage: Azure blob storage is Microsoft cloud storage. Blob storage can store a massive amount of file data as unstructured data. The unstructured data means not belong to any specific type, which means text or binary data. So something like images or pdf or videos to store in the cloud, then the most recommended is to use the blob store. The key component to creating azure blob storage resource: Storage Account:- A Storage account gives a unique namespace in Azure for all the data we will save. Every object that we store in Azure Storage has an address. The address is nothing but the unique name of our Storage Account name. The combination of the account name and the Azure Storage blob endpoint forms the base address for each object in our Storage account. For example, if our Storage Account is n

.Net5 Web API Redis Cache Using StackExchange.Redis.Extensions.AspNetCore Library

In this article, we are going to explore the integration of Redis cache in .Net5 Web API application using the 'StackExchange.Redis.Exntensions' library. Note:- Microsoft has introduced an 'IDistributedCache' interface in dotnet core which supports different cache stores like In-Memory, Redis, NCache, etc. It is simple and easy to work with  'IDistributedCache', for the Redis store with limited features but if we want more features of the Redis store we can choose to use 'StackExchange.Redis.Extensions'.  Click here for Redis Cache Integration Using IDistributedCache Interface . Overview On StackExchange.Redis.Extnesions Library: The 'StackExchange.Redis.Extension' library extended from the main library 'StackExchange.Redis'. Some of the key features of this library like: Default serialization and deserialization. Easy to save and fetch complex objects. Search key. Multiple Database Access Setup Redis Docker Instance: For this sampl

Endpoint Routing In Asp.Net Core

How Routing Works In  Core 2.1 And Below Versions?: In Asp.Net Core routing is configured using app.UseRouter() or app.UseMvc() middleware. app.UseMvc(routes => { routes.MapRoute( name: "default", template: "{controller=Home}/{action=Index}/{id?}"); }); Here in Dotnet Core version 2.1 or below versions on the execution of route middleware request will be navigated appropriate controller matched to the route. An operation or functionality which is dependent on route URL or route values and that need to be implemented before the execution of route middleware can be done by accessing the route path from the current request context as below app.Use(async (context, next) => { if(context.Request.Path.Value.IndexOf("oldvehicle") != -1) { context.Response.Redirect("vehicle"); } else { await next(); } }); app.UseMvc(routes => { routes.MapRoute( name: "vehicleRoute", template: "vehicle", defaul

Asp.Net Core MVC Form Validation Techniques

Introduction: Form validations in any applications are like assures that a valid data is storing on servers. All programing frameworks have their own individual implementations for form validations. In Dotnet Core MVC application server-side validations carried on by the models with the help of Data Annotations and the client-side validations carried by the plugin jQuery Unobtrusive Validation. jQuery Unobtrusive Validation is a custom library developed by Microsoft based on the popular library  jQuery Validate . In this article, we are going to learn how the model validation and client-side validation works in Asp.Net Core MVC Application with sample examples. Getting Started: Let's create an Asp.Net Core MVC application project using preferred editors like Microsoft Visual Studio or Microsoft Visual Studio Code. Here I'm using Visual Studio. Let's create an MVC controller and name it as 'PersonController.cs' and add an action method as bel

NestJS File Upload

In this article, we are going to understand the steps to create a file uploading endpoint in the NestJS application. Key Features In NestJS File Upload: Let us know some key features of NestJS file upload before implementing a sample application. FileInterceptor: The 'FileInterceptor' will be decorated on top of the file upload endpoint. This interceptor will read single file data from the form posted to the endpoint. export declare function FilesInterceptor(fieldName: string, localOptions?: MulterOptions): Type<NestInterceptor>; Here we can observe the 'fieldName' first input parameter this value should be a match with our 'name' attribute value on the form file input field. So our interceptor read our files that are attached to the file input field. Another input parameter of 'MulterOptions' that provides configuration like file destination path, customizing file name, etc. FilesInterceptor: The 'FilesInterceptor' will be decorated on t

.NET Core MVC Application File Upload To Physical Location With Buffered Technique

Buffering Technique In File Upload: The server will use its Memory(RAM) or Disk Storage to save the files on receiving a file upload request from the client.  Usage of Memory(RAM) or Disk depends on the number of file requests and the size of the file.  Any single buffered file exceeding 64KB is moved from Memory to a temp file on disk.  If an application receives heavy traffic of uploading files there might be a chance of out of Disk or RAM memory which leads to crash application. So this Buffer technique used for small files uploading. In the following article, we create a sample for the file uploading using .NET Core MVC application. Create The .NET Core MVC Project: Let's create a .NET Core MVC project, here for this sample I'm using Visual Studio Code as below.   Check the link to use the Visual Studio Code for .NET Core Application . IFormFile: Microsoft.AspNetCore.Http.IFormFile used for file upload with buffered technique. On uploading files f

Ionic Picker Sample Code In Angular

Introduction: Ionic Picker(ion-picker) is a popup slides up from the bottom of the device screen, which contains rows with selectable column separated items. The main building block of ion-picker as follows: PickerController PickerOptions PickerController: PickerController object helps in creating an ion-picker overlay. create(opts?: Opts): Promise<Overlay> PickerController create method helps in create the picker overlay with the picker options PickerOptions: PickerOptions is a configuration object used by PickerController to display ion-picker. Single Column Ionic Picker: single.item.picker.ts: import { Component } from "@angular/core"; import { PickerController } from "@ionic/angular"; import { PickerOptions } from "@ionic/core"; @Component({ selector: "single-column-picker", templateUrl:"single.item.picker.html" }) export class SingleItemPicker { animals: string[] = ["Tiger&quo

.Net Core HttpClient JSON Extension Methods Using System.Net.Http.Json Package

.Net Core 3.0 onwards Microsoft brought up a new package called System.Net.Http.Json. This new package provides JSON extension methods for HttpClient. These JSON extension methods will have a prebuild mechanism for serializing or deserializing response data or payload of HttpClient call. System.Net.Http.Json extension methods that are provided to HttpClient, few of them are mentioned below. GetFromJsonAsync PostAsJsonAsync PutAsJsonAsync ReadFromJsonAsync In this article, we understand System.Net.Http.Json package by implementing the HttpClient samples by with and without JSON extension methods and compare them. Create A .Net Core Web API Sample Application: Let's create a .Net Core sample Web API application, from this application we will consume another Web API by implementing HttpClient calls. We can create a Web API sample application using IDE like Visual Studio 2019(Supports .Net Core 3.0 plus) or  Visual Studio Code . Create A Typed Client: In .Net Core using the Http