Clickhouse: Test on query speed of A-share minute data [Part 2]

explain ：
1、my_db.stock_tb: It's a 22 Billion rows , about 15 Column and is a single table , The compressed space is 50G, The compression ratio is about 30 times ;
This table has been saved 2010 All the years since A Shares of 1 Minute line data . Why design a single table , I want to do a pressure test .
2、 I am on my home computer WSL2 Stand alone mode test under , Memory 32G.

Here is a simple test of the above table .

1、LIMIT N test

def get_data_from_ch(): # startdate : 2010-01-01
    
    client = Client('localhost')
    database_name = "my_db"
    table_name = 'stock_tb'
    code = "600036.XSHG"
    query_sql = f'SELECT * FROM {database_name}.{table_name} LIMIT 10000000'
    print(query_sql)
    data = client.execute(query_sql)
    return data
t0 = time.time()
data = get_data_from_ch()
t1 = time.time()
print(f"get_data cost time : {t1-t0} s! ")

Different tests were carried out LIMIT Number of tests ：

2、 Conditional query test
（1）、 Single condition

def get_data_from_ch(): # startdate : 2010-01-01
    
    client = Client('localhost')
    database_name = "my_db"
    table_name = 'stock_tb'
    code = "600036.XSHG"
    query_sql = f"SELECT * FROM {database_name}.{table_name} WHERE code = '{code}' LIMIT 10"
    print(query_sql)
    data = client.execute(query_sql)
    return data
t0 = time.time()
data = get_data_from_ch()
t1 = time.time()
print(f"get_data cost time : {t1-t0} s! ")

（2）、 Multi condition query of object and date

def get_data_from_ch_by_code_and_datetime():
    client = Client('localhost')
    database_name = "my_db"
    table_name = 'stock_tb'
    startdate = "2018-01-01"
    enddate = "2022-02-02"
    code = "600036.XSHG"
    query_sql = f"SELECT * FROM {database_name}.{table_name} WHERE code ='{code}' AND toDate(datetime) >= toDate('{startdate}') AND toDate(datetime) <= toDate('{enddate}') "
    print(query_sql)
    data = client.execute(query_sql)
    print(f"data : {len(data)}")
    return data

t0 = time.time()
data = get_data_from_ch_by_code_and_datetime()
t1 = time.time()
print(f"get_data cost time : {t1-t0} s! ")

Now look at , If we don't optimize the design , The reading speed is still not good , It can't meet the business , Especially the requirements of quantitative platform . In general ,100 Reading 10000 lines of data , exceed 1 Seconds is too slow .

Where is the optimization of design ？ Sub database and sub table ？ Asynchronous read ？