HBase PAL

HBase PAL or POJO Abstraction Layer is a library that abstracts retrieval and persistence of POJOs to HBase through the use of annotation that maps a POJO class to an HBase table and its fields to HBase columns.

Examples:

Sample Sales entity decorated with HBase and HBaseColumn

@HBase(tableName = "Sales")
@Data
public class Sales extends HBaseRow {

    @HBaseColumn
    private String region;
    @HBaseColumn
    private String country;
    @HBaseColumn
    private String itemType;
    @HBaseColumn
    private LocalDate orderDate;
    @HBaseColumn
    private int orderId;
    @HBaseColumn
    private int unitsSold;
    @HBaseColumn
    private double unitPrice;
    @HBaseColumn
    private double unitCost;
    
    @Override
    public byte[] getKey() {
        return Bytes.toBytes(orderId);
    }
}

Sample code to retrieve a list of Sales from HBase

try(HBaseClient hBaseClient = new HBaseClient(hBaseConfiguration)) {
    try(HBaseTable<Sales> salesHBaseTable = hBaseClient.getHBaseTable<>(Sales.class)) {
        Scan scan = new Scan();
        List<Sales> sales = salesHBaseTable.fetch(scan);
    }
}

Features:

Getting/Creating HBase Table

// references an HBase table represented by an entity
hBaseClient.getHBaseTable<>(Sales.class); 
// creates an HBase table based on the metadata specified in the entity
hBaseClient.createHBase(Sales.class);

Retrieval/persistence

// HBaseTable
List<T> fetch(Scan scan);   // get a list of type T from HBase table
Steam<T> stream(Scan scan); // get a stream of type T from HBase table
void refresh(List<T> rows); // refresh a list of type T with data from HBase table
void save(List<T> rows);    // saves a list of type T to HBase table

Row Key Salting

To support enterprise requirements to distribute rows across regions to avoid hot-spotting.

// if salt is specified, the internal salting algorithm
// will prefix the result of the getKey with a salt,
// salt value can only be between 1-255
@HBase(tableName = "Sales", salt = 8)
.
.
.
@Override
public byte[] getKey() {
    return Bytes.add(Bytes.toBytes(region.getValue()), Bytes.toBytes(orderId));
}

Custom column and family name mapping

@HbaseColumn(name = "columnName", family = "f")

Get/put enrichment and HBase result access

For additional columns you want to push/pull from HBase without mapping them to fields you can override the onAfterGet and onAfterPut methods. When you want access to the HBase result override onResult.

// HBaseRow Methods
void onGet(Get get);
void onPut(Put put);
void onResult(Result result);

Custom field converters

For cases where you have a field data type that cannot be directly converted to and from a binary array e.g. enumerations, you can extend HBaseColumnConverter to customize conversion of your field.

// Region is an enumeration and RegionColumnConvert converts the
// Region enumeration to  byte array and vice-versa
@HBaseColumn(converter = RegionColumnConverter.class)
private Region region;

Demo Code

The demo code assumes you have HBase installed on your local host or via Docker (I used dajobe/hbase personally, you can pull it directly from Docker Hub). I used IntelliJ for IDE, open up the demo project, it references the hbase-pal module.

Once you've setup your HBase, you need to run the demo application initially with a program argument of --spring.profiles.active=init for it to go through with the Sales table creation and data population (remember to remove the argument on subsequent runs). The sample includes a Swagger UI so you can test it by invoking http://localhost:8888/swagger-ui.html from your browser.

tuldoklambat / hbase-pal