Introduction
In the previous post we looked at table-related operations in Amazon DynamoDb: table creation, deletion and update. In this post we’ll continue our discussion of DynamoDb by insertions. We’ll see that there are two ways to build up the records and insert them into DynamoDb.
Open the demo application we’ve been working on in Visual Studio and let’s start.
Preparation
For this post we’ll need to recreate the test table we deleted at the end of the previous post. Make the following call through Main to create it anew under a different name:
DynamoDbDemoService service = new DynamoDbDemoService();
service.CreateNewTableDemo("People");
Wait until the table becomes active.
Inserting records into a DynamoDb table
There are two distinct ways to represent a record that should be inserted into DynamoDb through the AWS .NET API: the Document model and the Data model.
The Document model follows the traditional NoSql approach of loose coupling and loose typing. The properties of the object – which will be the record in the table – are described using strings. This is the model to follow in case you’d like to have the freedom to enter any record type into the table. Well, almost any record type, as you’ll need to provide values for the key(s) at a minimum.
The Data model follows OOP and strong typing. The properties of the record are described in proper classes and using DynamoDb specific attributes. This is the model to follow in case you want to apply a strict and predicable schema to your records.
Document model insertion
Let’s look at the document model first. Insert the following code into DynamoDbDemoService:
public void InsertPeopleByDocumentModel(string tableName)
{
try
{
using (IAmazonDynamoDB client = GetDynamoDbClient())
{
Table peopleTable = Table.LoadTable(client, tableName);
Document firstPerson = new Document();
firstPerson["Name"] = "John";
firstPerson["Birthdate"] = new DateTime(1980, 06, 24);
firstPerson["Address"] = "34 Wide Street, London, UK";
firstPerson["Age"] = 34;
firstPerson["Neighbours"] = new List<String>() { "Jane", "Samantha", "Richard" };
peopleTable.PutItem(firstPerson);
Document secondPerson = new Document();
secondPerson["Name"] = "Jill";
secondPerson["Birthdate"] = new DateTime(1981, 02, 26);
secondPerson["Address"] = "52 Broad Street, Dublin, Ireland";
secondPerson["Age"] = 33;
secondPerson["Neighbours"] = new List<String>() { "Alex", "Greg", "Michael" };
peopleTable.PutItem(secondPerson);
Document thirdPerson = new Document();
thirdPerson["Name"] = "George";
thirdPerson["Birthdate"] = new DateTime(1979, 11, 4);
thirdPerson["Address"] = "118 Main Street, Washington";
thirdPerson["Age"] = 35;
thirdPerson["Neighbours"] = new List<String>() { "Kathrine", "Kate", "Christine" };
peopleTable.PutItem(thirdPerson);
Document fourthPerson = new Document();
fourthPerson["Name"] = "Carole";
fourthPerson["Birthdate"] = new DateTime(1984, 4, 10);
fourthPerson["Address"] = "5 Short Street, Sydney, Australia";
fourthPerson["Age"] = 30;
fourthPerson["Neighbours"] = new List<String>() { "Nadia", "Katya", "Malcolm" };
peopleTable.PutItem(fourthPerson);
}
}
catch (AmazonDynamoDBException exception)
{
Debug.WriteLine(string.Concat("Exception while inserting records into DynamoDb table: {0}", exception.Message));
Debug.WriteLine(String.Concat("Error code: {0}, error type: {1}", exception.ErrorCode, exception.ErrorType));
}
}
We load the table and then build up 4 objects using a dictionary in the Document object. The objects describe 4 people that have some properties like Age and Address in common. You see that we use the general Document object and not some specialised Person object to do this. For each new record we call the PutItem method of the Table object.
Call the method from Main as follows:
DynamoDbDemoService service = new DynamoDbDemoService();
service.InsertPeopleByDocumentModel("People");
Highlight table in DynamoDb and click Explore table:
The records should be visible in our DynamoDb People table:
You’ll see that the birth dates look a bit funny as even the hour-minute-seconds sections were filled in, but it’s OK for now. The lists of neighbours were automatically converted into Json-like strings.
Delete all items from DynamoDb before we look at the Data model:
Data model
Here we’ll represent the loosely typed objects above as “proper” classes. Each Person object will have an Address object as well which will be formatted using Json.
Add a NuGet referencet to a JSON framework as Address object will be serialised and deserialised using the popular JSON.NET library:
Let’s insert the Address class first:
public class Address
{
public Address()
{
Street = "N/A";
City = "N/A";
Country = "N/A";
}
public string Street { get; set; }
public string City { get; set; }
public string Country { get; set; }
}
Each Person object will have an Address but the address must be represented in DynamoDb in one of the accepted data types: number, string or binary. The most straightforward solution is to convert an Address object into a Json string. The AWS .NET library has a special interface for that called IPropertyConverter which requires you to implement to methods: FromEntry and ToEntry. They help you describe how an object should be serialised and deserialised.
Insert the following address converter into the solution:
public class AddressConverter : IPropertyConverter
{
public object FromEntry(DynamoDBEntry entry)
{
Primitive primitive = entry as Primitive;
if (primitive == null) return new Address();
if (primitive.Type != DynamoDBEntryType.String)
{
throw new InvalidCastException(string.Format("Address cannot be converted as its type is {0} with a value of {1}"
, primitive.Type, primitive.Value));
}
string json = primitive.AsString();
return JsonConvert.DeserializeObject<Address>(json);
}
public DynamoDBEntry ToEntry(object value)
{
Address address = value as Address;
if (address == null) return null;
string json = JsonConvert.SerializeObject(address);
return new Primitive(json);
}
}
In FromEntry we check if the incoming DynamoDb entry is null and if it’s of type string. We can only convert strings into Address objects. We finally deserialise the JSON string into an Address object and return it. In ToEntry we do the opposite and and convert the incoming object first into an Address and then serialise it to its JSON representation.
We can now build the Person object:
[DynamoDBTable("People")]
public class Person
{
[DynamoDBHashKey]
public string Name { get; set; }
[DynamoDBRangeKey(AttributeName="Birthdate")]
public DateTime BirthDate { get; set; }
[DynamoDBProperty(Converter = typeof(AddressConverter))]
public Address Address { get; set; }
public int Age { get; set; }
public List<string> Neighbours { get; set; }
}
You can see how we use DynamoDb attributes to describe our hash and range keys. We also provide the converter for Address. You can use the DynamoDBProperty attribute to indicate in case the property name in DynamoDb is different from the property name in your class. This can be useful in case you follow a different naming structure in DynamoDb. Here’s an example:
[DynamoDBProperty(AttributeName="helloThisIsTheAge")]
public int Age { get; set; }
We can now insert the equivalent of the InsertPeopleByDocumentModel method using the Data model. Add the following InsertPeopleByDataModel method to DynamoDbDemoService:
[DynamoDBTable("People")]
public void InsertPeopleByDataModel(string tableName)
{
try
{
using (IAmazonDynamoDB client = GetDynamoDbClient())
{
DynamoDBContext context = new DynamoDBContext(client);
Person firstPerson = new Person()
{
Name = "John"
, BirthDate = new DateTime(1980, 06, 24)
, Address = new Address() { Street = "34 Wide Street", City = "London", Country = "UK" }
, Age = 34
, Neighbours = new List<String>() { "Jane", "Samantha", "Richard" }
};
Person secondPerson = new Person()
{
Name = "Jill"
, BirthDate = new DateTime(1981, 02, 26)
, Address = new Address() { Street = "52 Broad Street", City = "Dublin", Country = "Ireland" }
, Age = 33
, Neighbours = new List<String>() { "Alex", "Greg", "Michael" }
};
Person thirdPerson = new Person()
{
Name = "George"
, BirthDate = new DateTime(1979, 11, 4)
, Address = new Address() { Street = "118 Main Street", City = "Washington", Country = "USA" }
, Age = 35
, Neighbours = new List<String>() { "Kathrine", "Kate", "Christine" }
};
Person fourthPerson = new Person()
{
Name = "Carole"
, BirthDate = new DateTime(1984, 4, 10)
, Address = new Address() { Street = "5 Short Street", City = "Sydney", Country = "Australia" }
, Age = 30
, Neighbours = new List<String>() { "Nadia", "Katya", "Malcolm" }
};
context.Save<Person>(firstPerson);
context.Save<Person>(secondPerson);
context.Save<Person>(thirdPerson);
context.Save<Person>(fourthPerson);
}
}
catch (AmazonDynamoDBException exception)
{
Debug.WriteLine(string.Concat("Exception while inserting records into DynamoDb table: {0}", exception.Message));
Debug.WriteLine(String.Concat("Error code: {0}, error type: {1}", exception.ErrorCode, exception.ErrorType));
}
}
We build up the same 4 objects as before but now as proper Person objects. We finally use the DynamoDBContext object to save each new Person.
Call the method from Main:
service.InsertPeopleByDataModel("People");
You’ll find the records in our DynamoDb table:
Note how the Address is shown as Json.
In the next post we’ll see how to update and delete records using both models.
View all posts related to Amazon Web Services and Big Data here.
Exercises in .NET with Andras Nemes 