See http://mail-archives.apache.org/mod_mbox/pig-user/201101.mbox/%3CAA685952-71A7-4F4D-A1F6-E2C625696C06@syncapse.com%3E

dump_column.pig

a = LOAD 'test' USING org.apache.pig.backend.hadoop.hbase.HBaseStorage('data:column_1');
STORE a INTO 'out' USING PigStorage(',');  

hbase_test.py

import random
import sys

from org.apache.hadoop.hbase import HBaseConfiguration, HTableDescriptor, HColumnDescriptor
from org.apache.hadoop.hbase.client import HBaseAdmin, HTable, Put

def setup_table(name, *families):
    conf = HBaseConfiguration()
    desc = HTableDescriptor(name)
    [desc.addFamily(HColumnDescriptor('%s:' % f)) for f in families]

    admin = HBaseAdmin(conf)
    if admin.tableExists(name):
        admin.disableTable(name)
        admin.deleteTable(name)
    admin.createTable(desc)
    table = HTable(conf, name)
    return table

def put_row(table, row, **families):
    p = Put(row)
    for f, values in families.items():
        [p.add(f, k, v) for k,v in values.items()]
    return p

def data(cols):
    return dict([('column_%d' % i, 'value_%d_%d' % (i, random.randint(0, 5))) for i in range(cols)])

def rows(count, batch, **families):
    for i in xrange(0, count, batch):
        yield [put_row(t, 'row_%s' % (i+r), **families) for r in range(batch)]
        print '%s rows put' % (i + batch)

META = {'a':'b'}
if __name__ == '__main__':
    # Put n rows in batches of b
    name = sys.argv[1]
    cols, n, b = [int(p) for p in sys.argv[2:5]]

    t = setup_table(name, 'meta', 'data')
    [t.put(r) for r in rows(n, b, meta=META, data=data(cols))]