Cassandra Scan

gistfile1.py

import pycassa
import time
from pycassa.pool import ConnectionPool
from pycassa.columnfamily import ColumnFamily

servers = [
  '', 
]

pool = ConnectionPool('Keyspace1', servers, pool_size=1)
cf = ColumnFamily(pool, 'Standard1')

def main():
  total = 0
  batch = 5000
  
  last_key = ''
  elapsed_total = 0
  
  print 'last_key, total, elapsed, qps, qps_overall'

  while True:
    st = time.time()
    rows = cf.get_range(start=last_key, column_count=1, row_count=batch, read_consistency_level=pycassa.ConsistencyLevel.ONE)
    
    count = 0
    for row in rows:
      last_key = row[0]
      count += 1
    
    if count == 0:
      break
    
    total += count
    
    elapsed = time.time() - st
    elapsed_total += elapsed
    qps = batch / elapsed
    qps_overall = total / elapsed_total
    print last_key, total, elapsed, qps, qps_overall

main()