def test_indexer(self):
dictConfig(config.logging_config)
logger = logging.getLogger()
config.INDEX_CAPACITY = 1000000
table = Table("testdb", "test_table", "test_xcvzdfsadx", config,
logger)
store = table.store
indexer = table.indexer
max_range = 100000
insert_perf = []
key_list = []
for i in range(max_range):
key = ''.join(
random.choice(string.ascii_uppercase + string.digits)
for _ in range(10))
value = ''.join(
random.choice(string.ascii_uppercase + string.digits)
for _ in range(100))
expected_data = Record(key, value)
key_list.append(key)
position = store.save(expected_data)
indexer.put(expected_data.key, position, store)
print("Loading data progress: " + str(i * 100.0 / max_range) + "%",
end="\r")
print("\n total index files: " + str(len(indexer.index_list)))
total_seconds = 0.0
i = 0
for key in key_list:
start_time = timeit.default_timer()
indexer.get(key, store)
elapsed = timeit.default_timer() - start_time
insert_perf.append(elapsed * 1000.0) #to ms
total_seconds += elapsed
print("get test progress: " + str(i * 100.0 / max_range) + "%",
end="\r")
i += 1
plt.xlim([-1, max_range])
plt.ylim([0, 2])
plt.xlabel("get call")
plt.ylabel("ms")
plt.plot(insert_perf)
plt.title(COG_VERSION + " GET BECHMARK : " + str(max_range),
fontsize=12)
plt.savefig("get_bench.png")
print("\n ops/s: " + str(max_range / total_seconds))
print('\n num index files: ' + str(len(table.indexer.index_list)))
table.close()
def test_indexer_put_get(self):
if not os.path.exists("/tmp/" + DIR_NAME + "/test_table/"):
os.makedirs("/tmp/" + DIR_NAME + "/test_table/")
config.COG_HOME = DIR_NAME
print("*** " + config.COG_HOME + "\n")
dictConfig(config.logging_config)
logger = logging.getLogger()
table = Table("testdb", "test_table", "test_xcvzdfsadx", config,
logger)
store = table.store
indexer = table.indexer.index_list[0]
max_range = 100
for i in range(max_range):
key = ''.join(
random.choice(string.ascii_uppercase + string.digits)
for _ in range(10))
value = ''.join(
random.choice(string.ascii_uppercase + string.digits)
for _ in range(100))
expected_data = (key, value)
position = store.save(expected_data)
indexer.put(expected_data[0], position, store)
returned_data = indexer.get(expected_data[0], store)
print("indexer retrieved data: " + str(returned_data))
self.assertEqual(expected_data, returned_data[1])
print("Test progress: " + str(i * 100.0 / max_range))
c = 0
scanner = indexer.scanner(store)
for r in scanner:
# print r
c += 1
print("Total records scanned: " + str(c))
indexer.close()
store.close()
table.close()
def test_indexer(self):
dictConfig(config.logging_config)
config.INDEX_CAPACITY = 1000000
logger = logging.getLogger()
table = Table("testdb", "test_table", "test_xcvzdfsadx", config,
logger)
store = table.store
indexer = table.indexer
max_range = 100000
insert_perf = []
total_seconds = 0.0
for i in range(max_range):
key = ''.join(
random.choice(string.ascii_uppercase + string.digits)
for _ in range(10))
value = ''.join(
random.choice(string.ascii_uppercase + string.digits)
for _ in range(100))
expected_data = Record(key, value)
start_time = timeit.default_timer()
position = store.save(expected_data)
indexer.put(expected_data.key, position, store)
elapsed = timeit.default_timer() - start_time
insert_perf.append(elapsed * 1000.0) #to ms
total_seconds += elapsed
print("Test progress: " + str(i * 100.0 / max_range) + "%",
end="\r")
plt.xlim([-1, max_range])
plt.ylim([0, 2])
plt.xlabel("put call")
plt.ylabel("ms")
plt.plot(insert_perf)
plt.savefig("insert_bench.png")
print("\n total index files: " + str(len(indexer.index_list)))
print("\n ops/s: " + str(max_range / total_seconds))
table.close()
def test_indexer(self):
dictConfig(config.logging_config)
logger = logging.getLogger()
config.INDEX_CAPACITY = 10003
table = Table("testdb", "test_table", "test_xcvzdfsadx", config,
logger)
store = table.store
indexer = table.indexer
max_range = 100000
plt.title("CogDB v" + COG_VERSION + " BENCHMARK Total records:" +
str(max_range),
fontsize=12)
put_perf = []
key_list = []
total_seconds_put = 0.0
annotation = " index size: {}\n index_block_len: {}\n store read buffer: {}\n".format(
config.INDEX_CAPACITY, config.INDEX_BLOCK_LEN,
config.STORE_READ_BUFFER_SIZE)
print(annotation)
for i in range(max_range):
key = ''.join(
random.choice(string.ascii_uppercase + string.digits)
for _ in range(10))
value = ''.join(
random.choice(string.ascii_uppercase + string.digits)
for _ in range(100))
expected_data = Record(key, value)
key_list.append(key)
start_time = timeit.default_timer()
position = store.save(expected_data)
indexer.put(expected_data.key, position, store)
elapsed = timeit.default_timer() - start_time
put_perf.append(elapsed * 1000.0) # to ms
total_seconds_put += elapsed
print("Loading data progress: " + str(i * 100.0 / max_range) + "%",
end="\r")
plt.xlim([-1, max_range])
plt.ylim([0, 10])
plt.plot(put_perf, '-r', label="put")
annotation += "\n put ops/s: " + str(max_range / total_seconds_put)
total_index_file_size = 0
for i in table.indexer.index_list:
total_index_file_size += Path(i.name).stat().st_size
total_index_file_size = total_index_file_size >> 20
store_file_size = Path(table.store.store).stat().st_size >> 20
annotation += "\n index size: {}Mb \n store size: {}Mb ".format(
total_index_file_size, store_file_size)
get_perf = []
total_seconds_get = 0.0
i = 0
for key in key_list:
start_time = timeit.default_timer()
indexer.get(key, store)
elapsed = timeit.default_timer() - start_time
get_perf.append(elapsed * 1000.0) #to ms
total_seconds_get += elapsed
print("get test progress: " + str(i * 100.0 / max_range) + "%",
end="\r")
i += 1
plt.xlim([-1, max_range])
plt.ylim([0, 10])
plt.ylabel("ms")
plt.plot(get_perf, '-b', label='get')
plt.legend(loc="upper right")
annotation += "\n get ops/s: " + str(max_range / total_seconds_get)
annotation += '\n num index files: ' + str(
len(table.indexer.index_list))
plt.annotate(annotation, xy=(0.05, .5), xycoords='axes fraction')
notes_path = "../notes/"
if not os.path.exists(notes_path):
notes_path = "notes/"
plt.savefig("{}bench_{}.png".format(notes_path, max_range))
table.close()
