I have a server node based on the example mentioned in https://github.com/apache/ignite/blob/master/modules/platforms/dotnet/examples/ServerNode/Program.cs

Also in another process, I have a client with "ClientMode = true" like below.

        return new IgniteConfiguration(GetServerNodeConfiguration())
        {
            ClientMode = true
        };

The cache will hold the information of the "Employee" class, and I'm using IBinarySerializer to serialize the "Employee" and "Address" classes, and serializers are added into the IgniteConfiguration as below...

IgniteConfiguration.BinaryConfiguration = new BinaryConfiguration
                {
                    TypeConfigurations = new List<BinaryTypeConfiguration>
                    {
                        new(typeof(Employee)){Serializer = new EmployeeSerializer()},
                        new(typeof(Address)){Serializer = new AddressSerializer()},
                    }
                }

My questions are...

1. Do we need to have the "Employee" and "Address" classes both on the server and the client code? or do I just mention them on the client side?
2. Do we need to have the IBinarySerializer serializers in ignite BinaryConfiguration both on the server and the client code? or do I just mention them on the client-side?

Employee

public class Employee
{
    public Guid EmployeeId { get; set; }
    public string Name { get; set; }
    public Address Address { get; set; }
}

Address

public class Address
{
    public string Street { get; set; }
    public string Zip { get; set; }
}

IBinarySerializer implementations

public sealed class EmployeeSerializer : IBinarySerializer
{
    public void WriteBinary(object obj, IBinaryWriter writer)
    {
        var employee = (Employee)obj;

        writer.WriteGuid("employeeId", employee.EmployeeId );
        writer.WriteString("name", employee.Name);
        writer.WriteObject("address", employee.Address);
    }

    public void ReadBinary(object obj, IBinaryReader reader)
    {
        var employee = (employee)obj;

        employee.EmployeeId  = reader.ReadObject<Guid>("employeeId");
        employee.Name = reader.ReadString("name");
        employee.Address = reader.ReadObject<Address>("address");
    }
}


public sealed class AddressSerializer : IBinarySerializer
{
    public void WriteBinary(object obj, IBinaryWriter writer)
    {
        var address = (Address)obj;

        writer.WriteString("street", address.Street);
        writer.WriteString("zip", address.Zip);
    }

    public void ReadBinary(object obj, IBinaryReader reader)
    {
        var address = (Address)obj;

        address.Street = reader.ReadString("street");
        address.Zip = reader.ReadString("zip");
    }
}

CodePudding user response：

In short, yes, you need this information on each node. Both sides have to know how to deserialize and serialize a record properly.

Though it's possible to work directly in binary mode using BinaryObjectBuilder without explicit models definition. It's not mandatory to define a custom IBinarySerializer either, the internal serializer will do the trick well enough unless you have some big models with dozens of fields defined.