Getting Started with LINQ in the ScalarDB Cluster .NET Client SDK

The ScalarDB Cluster .NET Client SDK supports querying the cluster with LINQ and some Entity Framework-like functionality.

注記

This SDK doesn't support Entity Framework. Instead, this SDK implements functionality that is similar to Entity Framework.

注記

SQL support must be enabled on the cluster to use LINQ.

Set up classes

After confirming that SQL support is enabled, create a C# class for each ScalarDB table that you want to use. For example:

using System.ComponentModel.DataAnnotations.Schema;
using ScalarDB.Net.Client.DataAnnotations;

// ...

[Table("ns.statements")]
public class Statement
{
    [Column("statement_id", Order = 0), PartitionKey]
    public int Id { get; set; }
    
    [Column("order_id", Order = 1), SecondaryIndex]
    public string OrderId { get; set; } = String.Empty;

    [Column("item_id", Order = 2), SecondaryIndex]
    public int ItemId { get; set; }

    [Column("count", Order = 3)]
    public int Count { get; set; }
}

[Table("order_service.items")]
public class Item
{
    [Column("item_id", Order = 0), PartitionKey]
    public int Id { get; set; }

    [Column("name", Order = 1)]
    public string Name { get; set; } = String.Empty;

    [Column("price", Order = 2)]
    public int Price { get; set; }
}

If a partition key, clustering key, or secondary index consists of more than one column, the Order property of ColumnAttribute will decide the order inside the key or index.

Create a context class that has properties for all the tables you want to use. For example:

    public class MyDbContext: ScalarDbContext
    {
        public ScalarDbSet<Statement> Statements { get; set; }
        public ScalarDbSet<Item> Items { get; set; }
    }

After all the classes are created, you need to add the created context to the Dependency Injection. For example:

using ScalarDB.Net.Client.Extensions;

//...

var builder = WebApplication.CreateBuilder(args);

//...

builder.Services.AddScalarDbContext<MyDbContext>(options =>
{
    options.Address = "http://<HOSTNAME_OR_IP_ADDRESS>:<PORT>";
    options.HopLimit = 10;
});

The context can be injected into the controller's constructor as follows:

[ApiController]
public class OrderController: ControllerBase
{
    private readonly MyDbContext _myDbContext;

    public OrderController(MyDbContext myDbContext)
    {
        _myDbContext = myDbContext;
    }
}

Use LINQ to query properties

After receiving MyDbContext in your controller, you can query its properties by using LINQ. For example:

Use query syntax

from stat in _myDbContext.Statements
join item in _myDbContext.Items on stat.ItemId equals item.Id
where stat.Count > 2 && item.Name.Contains("apple")
orderby stat.Count descending, stat.ItemId
select new { item.Name, stat.Count };

Use method syntax

_myDbContext.Statements
            .Where(stat => stat.OrderId == "1")
            .Skip(1)
            .Take(2);

Use the First method to get one Statement by its partition key

_myDbContext.Statements.First(stat => stat.OrderId == "1");

Use the DefaultIfEmpty method to perform left outer join

from stat in _myDbContext.Statements
join item in _myDbContext.Items on stat.ItemId equals item.Id into items
from i in items.DefaultIfEmpty()
select new { ItemName = i != null ? i.Name : "" }

The following methods are supported:

• Select
• Where
• Join
• GroupJoin
• First/FirstOrDefault
• Skip
• Take
• OrderBy/OrderByDescending
• ThenBy/ThenByDescending

The following String methods are supported inside the predicates of Where and First/FirstOrDefault methods:

• Contains
• StartsWith
• EndsWith

Unsupported LINQ methods can be used after the supported methods. For example:

_myDbContext.Statements
            .Where(stat => stat.OrderId == "1") // Will be executed remotely on the cluster.
            .Distinct() // Will be executed locally in the app.
            .Where(stat => stat.ItemId < 5); // Will be executed locally.

注記

If Skip is specified before Take or First/FirstOrDefault, the number that is passed to Skip will be added to the LIMIT number in the SQL query. By itself, Skip won't change the resulting SQL query.

Limitations when using LINQ against ScalarDbSet{T} objects

• All method calls are supported inside Select. For example:

.Select(stat => convertToSomething(stat.ItemId))
//...
.Select(stat => stat.ItemId * getSomeNumber())

• Method calls, except for calls against the querying object, are also supported inside Where and First/FirstOrDefault. For example:

.Where(stat => stat.ItemId == getItemId()) // is OK
//...
.Where(stat => stat.ItemId.ToString() == "1") // is not supported

• All method calls are supported inside the result-selecting lambda of Join and GroupJoin. For example:

.Join(_myDbContext.Items,
      stat => stat.ItemId,
      item => item.Id,
      (stat, item) => new { ItemName = convertToSomething(item.Name),
                            ItemCount = stat.Count.ToString() })

• Method calls are not supported inside the key-selecting lambdas of Join and GroupJoin.
• Custom equality comparers are not supported. The comparer argument in Join and GroupJoin methods will be ignored if the argument has been passed.
• More than one from directly in one query is not supported, except when the DefaultIfEmpty method is used to perform left outer join. Each subsequent from is considered to be a separate query.

var firstQuery = from stat in _myDbContext.Statements
                 where stat.Count > 2
                 select new { stat.Count };

var secondQuery = from item in _myDbContext.Items
                  where item.Price > 6
                  select new { item.Name };

var finalQuery = from first in firstQuery
                 from second in secondQuery
                 select new { first.Count, second.Name };

// 1. firstQuery will be executed against the cluster.
// 2. secondQuery will be executed against the cluster for each object (row) from 1.
// 3. finalQuery will be executed locally with the results from 1 and 2.
var result = finalQuery.ToArray();

• Method calls are not supported inside OrderBy/OrderByDescending or ThenBy/ThenByDescending.
• Only overloads of Contains, StartsWith, and EndsWith methods that have a single string argument are supported inside Where and First/FirstOrDefault.

Modify data in a cluster by using ScalarDbContext

The properties of the class inherited from ScalarDbContext can be used to modify data.

Add a new object by using the AddAsync method

var statement = new Statement
                {
                    OrderId = "2",
                    ItemId = 4,
                    Count = 8
                };
await _myDbContext.Statements.AddAsync(statement);

Update an object by using the UpdateAsync method

var statement = _myDbContext.Statements.First(stat => stat.Id == 1);

// ...

statement.Count = 10;        
await _myDbContext.Statements.UpdateAsync(statement);

Remove an object by using the RemoveAsync method

var statement = _myDbContext.Statements.First(stat => stat.Id == 1);

// ...
    
await _myDbContext.Statements.RemoveAsync(statement);

Manage transactions

LINQ queries and AddAsync, UpdateAsync, and RemoveAsync methods can be executed without an explicitly started transaction. However, to execute multiple queries and methods as part of a single transaction, the transaction must be explicitly started and committed. ScalarDbContext supports both ordinary transactions and transactions with the two-phase commit interface in ScalarDB.

Begin a new transaction

await _myDbContext.BeginTransactionAsync();

Begin a new transaction with the two-phase commit interface

using ScalarDB.Net.Client.Core;

// ...

await _myDbContext.BeginTransactionAsync(TransactionType.TwoPhaseCommit);

Get the ID of a currently active transaction

var transactionId = _myDbContext.CurrentTransactionId;

Get the type of a currently active transaction

var transactionType = _myDbContext.CurrentTransactionType;

Join an existing transaction with the two-phase commit interface

using ScalarDB.Net.Client.Core;

// ...

await _myDbContext.JoinTransactionAsync(transactionId, TransactionType.TwoPhaseCommit);

Resume an existing transaction

await _myDbContext.ResumeTransaction(transactionId);

Resume an existing transaction with the two-phase commit interface

await _myDbContext.ResumeTransaction(transactionId, TransactionType.TwoPhaseCommit);

注記

The ResumeTransaction method doesn't have an asynchronous version because it only initializes the transaction data in the ScalarDbContext inheriting object without querying the cluster. Because of this, resuming a transaction by using the wrong ID is possible.

Commit a transaction (ordinary or two-phase commit)

await _myDbContext.CommitTransactionAsync();

Roll back a transaction (ordinary or two-phase commit)

await _myDbContext.RollbackTransactionAsync();

Prepare a transaction with the two-phase commit interface for the commit

await _myDbContext.PrepareTransactionAsync();

Validate a transaction with the two-phase commit interface before the commit

await _myDbContext.ValidateTransactionAsync();