HBase · bigdata Docs

A small PHP client for Apache HBase over its REST API (Stargate, the HBase REST server). It speaks JSON and handles the Base64 encoding of row keys, columns and values for you.

Requires a running HBase REST server, usually started with:

hbase rest start -p 8080


Driver name	`hbase`
Client	`Erikwang2013\BigData\Driver\HBase\HBaseClient`
Config	`Erikwang2013\BigData\Driver\HBase\HBaseConfig`
Optional dependency	none

Quick start

use Erikwang2013\BigData\BigData;

$hbase = BigData::driver('hbase', [
    'baseUrl' => 'http://127.0.0.1:8080',
]);

$hbase->createTable('users', ['info']);

$hbase->put('users', 'row1', [
    'info:name' => 'alice',
    'info:email' => 'alice@example.com',
]);

$cells = $hbase->get('users', 'row1');
// ['info:name' => 'alice', 'info:email' => 'alice@example.com']

$hbase->delete('users', 'row1');
$hbase->deleteTable('users');

Direct instantiation:

use Erikwang2013\BigData\Driver\HBase\HBaseClient;
use Erikwang2013\BigData\Driver\HBase\HBaseConfig;

$hbase = new HBaseClient(HBaseConfig::fromArray([
    'baseUrl' => 'http://127.0.0.1:8080',
]));
// Optional HTTP client and PSR-3 logger: new HBaseClient($config, $http, $logger)

Configuration

HBaseConfig::fromArray() accepts:

Key	Default	Description
`baseUrl`	required	REST server URL, e.g. `http://127.0.0.1:8080`.
`username`	`null`	Username for HTTP Basic auth.
`password`	`null`	Password for HTTP Basic auth.
`verify_ssl`	`true`	Verify TLS certificates.
`timeout`	`30.0`	Request timeout in seconds.

base_url is accepted as an alias for baseUrl.

API

Method	Description
`tables(): array`	List table names.
`createTable(string $table, array $columns): void`	Create a table; `$columns` is a list of column family names.
`deleteTable(string $table): void`	Delete a table and its data.
`get(string $table, string $row, array $columns = []): array`	Fetch a row as `column => value`; `$columns` optionally restricts to `family:qualifier` columns.
`put(string $table, string $row, array $cells): void`	Write `column => value` cells to a row.
`delete(string $table, string $row): void`	Delete a row.
`scan(string $table, array $options = []): array`	Create a scanner, fetch one batch and delete it.
`scannerCreate(string $table, array $options = []): string`	Create a scanner; returns its path (e.g. `/users/scanner/42`).
`scannerGet(string $scannerId, int $count = 100): array`	Fetch the next batch; `[]` once the scanner is exhausted.
`scannerDelete(string $scannerId): void`	Delete a scanner.

scan() and scannerGet() return a list of rows:

[
    ['key' => 'row1', 'cells' => ['info:name' => 'alice']],
    ['key' => 'row2', 'cells' => ['info:name' => 'bob']],
]

Scan options: startRow, endRow (row keys are encoded for you) and limit (batch size requested from the scanner, default 100). A scan fetches a single batch of up to limit rows; use scannerCreate() / scannerGet() to page through larger tables.

Scanner ids returned by scannerCreate() include the table path, so pass the value straight to scannerGet() and scannerDelete().

Errors

Non-2xx responses throw Erikwang2013\BigData\Exception\RequestException with the status code and response body ($statusCode, $body).
Network and timeout failures throw Erikwang2013\BigData\Exception\ConnectionException.
Both extend Erikwang2013\BigData\Exception\BigDataException.

Notes

Requires the HBase REST (Stargate) server; it does not speak the native HBase RPC protocol.
Scan pagination across batches is manual via the scanner methods.
No automatic retries are performed.