On this post we will create Tables on a DynamoDB Database the java way.
Before getting started we need to have local dynamodb installed since we want to avoid any costs for dynamodb usage. There was a previous post on local dynamodb.
In case you use docker you can find a local dynamodb image or you can create one on you own as described here.
The dynamodb java sdk gives us the ability to create dynamodb tables using java code.
The most basic action is to create a table with a hash key. In this case the email of the user would be the hash key.
List<KeySchemaElement> elements = new ArrayList<KeySchemaElement>();
KeySchemaElement keySchemaElement = new KeySchemaElement()
.withKeyType(KeyType.HASH)
.withAttributeName("email");
elements.add(keySchemaElement);
List<AttributeDefinition> attributeDefinitions = new ArrayList<>();
attributeDefinitions.add(new AttributeDefinition()
.withAttributeName("email")
.withAttributeType(ScalarAttributeType.S));
CreateTableRequest createTableRequest = new CreateTableRequest()
.withTableName("Users")
.withKeySchema(elements)
.withProvisionedThroughput(new ProvisionedThroughput()
.withReadCapacityUnits(5L)
.withWriteCapacityUnits(5L))
.withAttributeDefinitions(attributeDefinitions);
amazonDynamoDB.createTable(createTableRequest);
What we did is creating the Users table using his email for a hash key.
The next table will be called Logins. Logins should keep track each time the user logged in. To do so apart from using a hash key we will also use a range key.
List<KeySchemaElement> elements = new ArrayList<KeySchemaElement>();
KeySchemaElement hashKey = new KeySchemaElement()
.withKeyType(KeyType.HASH)
.withAttributeName("email");
KeySchemaElement rangeKey = new KeySchemaElement()
.withKeyType(KeyType.RANGE)
.withAttributeName("timestamp");
elements.add(hashKey);
elements.add(rangeKey);
List<AttributeDefinition> attributeDefinitions = new ArrayList<>();
attributeDefinitions.add(new AttributeDefinition()
.withAttributeName("email")
.withAttributeType(ScalarAttributeType.S));
attributeDefinitions.add(new AttributeDefinition()
.withAttributeName("timestamp")
.withAttributeType(ScalarAttributeType.N));
CreateTableRequest createTableRequest = new CreateTableRequest()
.withTableName("Logins")
.withKeySchema(elements)
.withProvisionedThroughput(new ProvisionedThroughput()
.withReadCapacityUnits(5L)
.withWriteCapacityUnits(5L))
.withAttributeDefinitions(attributeDefinitions);
amazonDynamoDB.createTable(createTableRequest);
By using the email as a hash key we can query for the logins of the specific user.
By using the date that the login occured as a range key with can find sort the login entries or perform advanced queries based on the login date for a specific user.
However most of the times a hash key and range key are not enough for our needs.
DynamoDB provides us with Global Secondary indexes and Local secondary Indexes.
We will create the table SupervisorS. The hash key of Supervisor would be his name. A supervisor will work for a company. The company will be our global secondary index. Since the companies own more than one factories the field factory would be the range key.
List<KeySchemaElement> elements = new ArrayList<>();
KeySchemaElement hashKey = new KeySchemaElement()
.withKeyType(KeyType.HASH)
.withAttributeName("name");
elements.add(hashKey);
List<GlobalSecondaryIndex> globalSecondaryIndices = new ArrayList<>();
ArrayList<KeySchemaElement> indexKeySchema = new ArrayList<>();
indexKeySchema.add(new KeySchemaElement()
.withAttributeName("company")
.withKeyType(KeyType.HASH)); //Partition key
indexKeySchema.add(new KeySchemaElement()
.withAttributeName("factory")
.withKeyType(KeyType.RANGE)); //Sort key
GlobalSecondaryIndex factoryIndex = new GlobalSecondaryIndex()
.withIndexName("FactoryIndex")
.withProvisionedThroughput(new ProvisionedThroughput()
.withReadCapacityUnits((long) 10)
.withWriteCapacityUnits((long) 1))
.withKeySchema(indexKeySchema)
.withProjection(new Projection().withProjectionType(ProjectionType.ALL));
globalSecondaryIndices.add(factoryIndex);
List<AttributeDefinition> attributeDefinitions = new ArrayList<>();
attributeDefinitions.add(new AttributeDefinition()
.withAttributeName("name")
.withAttributeType(ScalarAttributeType.S));
attributeDefinitions.add(new AttributeDefinition()
.withAttributeName("company")
.withAttributeType(ScalarAttributeType.S));
attributeDefinitions.add(new AttributeDefinition()
.withAttributeName("factory")
.withAttributeType(ScalarAttributeType.S));
CreateTableRequest createTableRequest = new CreateTableRequest()
.withTableName("Supervisors")
.withKeySchema(elements)
.withProvisionedThroughput(new ProvisionedThroughput()
.withReadCapacityUnits(5L)
.withWriteCapacityUnits(5L))
.withGlobalSecondaryIndexes(factoryIndex)
.withAttributeDefinitions(attributeDefinitions);
amazonDynamoDB.createTable(createTableRequest);
Next table would be the table Companies. The hash key would be the parent company and the range key the subsidiary company. Each company has a CEO. The CEO would be the range key for the local secondary index.
List<KeySchemaElement> elements = new ArrayList<>();
KeySchemaElement hashKey = new KeySchemaElement()
.withKeyType(KeyType.HASH)
.withAttributeName("name");
KeySchemaElement rangeKey = new KeySchemaElement()
.withKeyType(KeyType.RANGE)
.withAttributeName("subsidiary");
elements.add(hashKey);
elements.add(rangeKey);
List<LocalSecondaryIndex> localSecondaryIndices = new ArrayList<>();
ArrayList<KeySchemaElement> indexKeySchema = new ArrayList<>();
indexKeySchema.add(new KeySchemaElement()
.withAttributeName("name")
.withKeyType(KeyType.HASH));
indexKeySchema.add(new KeySchemaElement()
.withAttributeName("ceo")
.withKeyType(KeyType.RANGE));
LocalSecondaryIndex ceoIndex = new LocalSecondaryIndex()
.withIndexName("CeoIndex")
.withKeySchema(indexKeySchema)
.withProjection(new Projection().withProjectionType(ProjectionType.ALL));
localSecondaryIndices.add(ceoIndex);
List<AttributeDefinition> attributeDefinitions = new ArrayList<>();
attributeDefinitions.add(new AttributeDefinition()
.withAttributeName("name")
.withAttributeType(ScalarAttributeType.S));
attributeDefinitions.add(new AttributeDefinition()
.withAttributeName("subsidiary")
.withAttributeType(ScalarAttributeType.S));
attributeDefinitions.add(new AttributeDefinition()
.withAttributeName("ceo")
.withAttributeType(ScalarAttributeType.S));
CreateTableRequest createTableRequest = new CreateTableRequest()
.withTableName("Companies")
.withKeySchema(elements)
.withProvisionedThroughput(new ProvisionedThroughput()
.withReadCapacityUnits(5L)
.withWriteCapacityUnits(5L))
.withLocalSecondaryIndexes(localSecondaryIndices)
.withAttributeDefinitions(attributeDefinitions);
amazonDynamoDB.createTable(createTableRequest);
You can find the source code on github.
I’ve compiled a cheat sheet that lists dynamodb functions in Java
Sign up in the link to receive it.
Emmanouil Gkatziouras 
Hi,
Thanks for this… you provide great examples of how to create DynamoDB tables, but only with scalar attributes.
However, I’m trying to set up a simple GIS database, where each feature has some scalar attributes (name, etc), but also a list of X,Y coordinate pairs. It appears that DynamoDB supports attributes containing lists, but I can’t see how to create one (outside of the AWS’s GUI). Can you please add a page on how to do complex datatypes? (I know I could do it by making the coordinate list a string (i.e. a scalar), but would rather “do it right”.
I think you need to add a converter that maps a string to a list.
After a day of experimenting, the key idea is that you DON’T have to declare all your attributes and their types when you create a table. You only have to declare attributes that are used in indexes. When I think about it, this makes sense. Indexes need to know what types of data they’re storing, so need attributes to be defined. Other attributes are stored as key-value pairs, and new keys can be created on the fly, so they don’t need to be predefined.
So I can create a table without even mentioning that I will have a list of coordinates in every item. I can then insert items containing the coordinates list by using DynamoDBMapper.
As you suggested, I did need to create a DynamoDBTypeConverter<String, List> class to convert the list items to strings.
That’s great! Eventually you might need to work your way with the sorting key and geolocation if you want to do some smart queries. One familiar challenge on that was back when I was using GAE where only one conditional was possible. I think Geohash might be useful to you https://read.acloud.guru/location-based-search-results-with-dynamodb-and-geohash-267727e5d54f
One quick addition to my previous message… when creating a table you can ONLY declare attributes which are used in indexes.
I’m fortunate in that I’m working on the NZ map grid, which means I don’t have to deal with lat/lon or great circles… everything is on a 2D metric grid. My intention was to use secondary indexes on the north, south, east, west bounds of each object for quick searching. But the single-index problem was lurking at the edge of my perception. Thanks for the geohash idea… it’s a neat solution, and I’ll build something like that.