Here are some results from a little testing I recent did on extracting data from an Oracle database and what effect the arraysize makes and which method might be the quickest.
The arraysize determines how many records will be retrieved in each each batch. When a query is issued to the database, the results are returned to the calling programme in batches of a certain size. Depending on the nature of the application and the number of records being retrieved, will determine the arraysize value. The value of this can have a dramatic effect on your query and application response times. Sometimes a small value works very well but sometimes you might need a larger value.
My test involved using an Oracle Database Cloud instance, using Python and the following values for the arraysize.
arraysize = (5, 50, 500, 1000, 2000, 3000, 4000, 5000)
The first test was to see what effect these arraysizes have on retrieving all the data from a table. The in question has 73,668 records. So not a large table. The test loops through this list of values and fetches all the data, using the fetchall function (part of cx_Oracle), and then displays the time taken to retrieve the results.
# import the Oracle Python library import cx_Oracle import datetime import pandas as pd import numpy as np # setting display width for outputs in PyCharm desired_width = 280 pd.set_option('display.width', desired_width) np.set_printoptions(linewidth=desired_width) pd.set_option('display.max_columns',30) # define the login details p_username = "************" p_password = "************" p_host = "************" p_service = "************" p_port = "1521" print('--------------------------------------------------------------------------') print(' Testing the time to extract data from an Oracle Database.') print(' using different approaches.') print('---') # create the connection con = cx_Oracle.connect(user=p_username, password=p_password, dsn=p_host+"/"+p_service+":"+p_port) print('') print(' Test 1: Extracting data using Cursor for different Array sizes') print(' Array Size = 5, 50, 500, 1000, 2000, 3000, 4000, 5000') print('') print(' Starting test at : ', datetime.datetime.now()) beginTime = datetime.datetime.now() cur_array_size = (5, 50, 500, 1000, 2000, 3000, 4000, 5000) sql = 'select * from banking_marketing_data_balance_v' for size in cur_array_size: startTime = datetime.datetime.now() cur = con.cursor() cur.arraysize = size results = cur.execute(sql).fetchall() print(' Time taken : array size = ', size, ' = ', datetime.datetime.now()-startTime, ' seconds, num of records = ', len(results)) cur.close() print('') print(' Test 1: Time take = ', datetime.datetime.now()-beginTime) print('')
And here are the results from this first test.
Starting test at : 2018-11-14 15:51:15.530002 Time taken : array size = 5 = 0:36:31.855690 seconds, num of records = 73668 Time taken : array size = 50 = 0:05:32.444967 seconds, num of records = 73668 Time taken : array size = 500 = 0:00:40.757931 seconds, num of records = 73668 Time taken : array size = 1000 = 0:00:14.306910 seconds, num of records = 73668 Time taken : array size = 2000 = 0:00:10.182356 seconds, num of records = 73668 Time taken : array size = 3000 = 0:00:20.894687 seconds, num of records = 73668 Time taken : array size = 4000 = 0:00:07.843796 seconds, num of records = 73668 Time taken : array size = 5000 = 0:00:06.242697 seconds, num of records = 73668
As you can see the variation in the results.
You may get different performance results based on your location, network connectivity and proximity of the database. I was at home (Ireland) using wifi and my database was located somewhere in USA. I ran the rest a number of times and the timings varied by +/- 15%, which is a lot!
When the data is retrieved in this manner you can process the data set in the returned results set. Or what is more traditional you will want to work with the data set as a panda. The next two test look at a couple of methods of querying the data and storing the result sets in a panda.
For these two test, I’ll set the arraysize = 3000. Let’s see what happens.
For the second test I’ll again use the fetchall() function to retrieve the data set. From that I extract the names of the columns and then create a panda combining the results data set and the column names.
startTime = datetime.datetime.now() print(' Starting test at : ', startTime) cur = con.cursor() cur.arraysize = cur_array_size results = cur.execute(sql).fetchall() print(' Fetched ', len(results), ' in ', datetime.datetime.now()-startTime, ' seconds at ', datetime.datetime.now()) startTime2 = datetime.datetime.now() col_names =  for i in range(0, len(cur.description)): col_names.append(cur.description[i])
print(' Fetched data & Created the list of Column names in ', datetime.datetime.now()-startTime, ' seconds at ', datetime.datetime.now())
The results from this are.
Fetched 73668 in 0:00:07.778850 seconds at 2018-11-14 16:35:07.840910 Fetched data & Created the list of Column names in 0:00:07.779043 seconds at 2018-11-14 16:35:07.841093 Finished creating Dataframe in 0:00:07.975074 seconds at 2018-11-14 16:35:08.037134 Test 2: Total Time take = 0:00:07.975614
Now that was quick. Fetching the data set in just over 7.7788 seconds. Creating the column names as fractions of a millisecond, and then the final creation of the panda took approx 0.13 seconds.
For the third these I used the pandas library function called read_sql(). This function takes two inputs. The first is the query to be processed and the second the name of the database connection.
print(' Test 3: Test timing for read_sql into a dataframe') cur_array_size = 3000 print(' will use arraysize = ', cur_array_size) print('') startTime = datetime.datetime.now() print(' Starting test at : ', startTime) df2 = pd.read_sql(sql, con) print(' Finished creating Dataframe in ', datetime.datetime.now()-startTime, ' seconds at ', datetime.datetime.now()) # close the connection at end of experiments con.close()
and the results from this are.
Test 3: Test timing for read_sql into a dataframe will use arraysize = 3000 Starting test at : 2018-11-14 16:35:08.095189 Finished creating Dataframe in 0:02:03.200411 seconds at 2018-11-14 16:37:11.295611
You can see that it took just over 2 minutes to create the panda data frame using the read_sql() function, compared to just under 8 seconds using the previous method.
It is important to test the various options for processing your data and find the one that works best in your environment. As with most languages there can be many ways to do the same thing. The challenge is to work out which one you should use.
If you used other languages, including Oracle PL/SQL, more than likely you will have experienced having to play buffering the number of records that are returned from a cursor. Typically this is needed when you are processing more than a few hundred records. The default buffering size is relatively small and by increasing the size of the number of records to be buffered can dramatically improve the performance of your code.
As with all things in coding and IT, the phrase “It Depends” applies here and changing the buffering size may not be what you need and my not help you to gain optimal performance for your code.
There are lots and lots of examples of how to test this in PL/SQL and other languages, but what I’m going to show you here in this blog post is to change the buffering size when using Python to process data in an Oracle Database using the Oracle Python library cx_Oracle.
Let us begin with taking the defaults and seeing what happens. In this first scenario the default buffering is used. Here we execute a query and the process the records in a FOR loop (yes these is a row-by-row, slow-by-slow approach.
import time i = 0 # define a cursor to use with the connection cur2 = con.cursor() # execute a query returning the results to the cursor print("Starting cursor at", time.ctime()) cur2.execute('select * from sh.customers') print("Finished cursor at", time.ctime()) # for each row returned to the cursor, print the record print("Starting for loop", time.ctime()) t0 = time.time() for row in cur2: i = i+1 if (i%10000) == 0: print(i,"records processed", time.ctime()) t1 = time.time() print("Finished for loop at", time.ctime()) print("Number of records counted = ", i) ttime = t1 - t0 print("in ", ttime, "seconds.")
This gives us the following output.
Starting cursor at 10:11:43 Finished cursor at 10:11:43 Starting for loop 10:11:43 10000 records processed 10:11:49 20000 records processed 10:11:54 30000 records processed 10:11:59 40000 records processed 10:12:05 50000 records processed 10:12:09 Finished for loop at 10:12:11 Number of records counted = 55500 in 28.398550033569336 seconds.
Processing the data this way takes approx. 28 seconds and this corresponds to the buffering of approx 50-75 records at a time. This involves many, many, many round trips to the the database to retrieve this data. This default processing might be fine when our query is only retrieving a small number of records, but as our data set or results set from the query increases so does the time it takes to process the query.
But we have a simple way of reducing the time taken, as the number of records in our results set increases. We can do this by increasing the number of records that are buffered. This can be done by changing the size of the ‘arrysize’ for the cursor definition. This reduces the number of “roundtrips” made to the database, often reducing networks load and reducing the number of context switches on the database server.
The following gives an example of same code with one additional line.
cur2.arraysize = 500
Here is the full code example.
# Test : Change the arraysize and see what impact that has import time i = 0 # define a cursor to use with the connection cur2 = con.cursor() cur2.arraysize = 500 # execute a query returning the results to the cursor print("Starting cursor at", time.ctime()) cur2.execute('select * from sh.customers') print("Finished cursor at", time.ctime()) # for each row returned to the cursor, print the record print("Starting for loop", time.ctime()) t0 = time.time() for row in cur2: i = i+1 if (i%10000) == 0: print(i,"records processed", time.ctime()) t1 = time.time() print("Finished for loop at", time.ctime()) print("Number of records counted = ", i) ttime = t1 - t0 print("in ", ttime, "seconds.")
Now the response time to process all the records is.
Starting cursor at 10:13:02
Finished cursor at 10:13:02
Starting for loop 10:13:02
10000 records processed 10:13:04
20000 records processed 10:13:06
30000 records processed 10:13:08
40000 records processed 10:13:10
50000 records processed 10:13:12
Finished for loop at 10:13:13
Number of records counted = 55500
in 11.780734777450562 seconds.
All done in just under 12 seconds, compared to 28 seconds previously.
Here is another alternative way of processing the data and retrieves the entire results set, using the ‘fetchall’ command, and stores it located in ‘res’.
Is Python the new R?
Maybe, maybe not, but that I’m finding in recent months is more companies are asking me to use Python instead of R for some of my work.
In this blog post I will walk through the steps of setting up the Oracle driver for Python, called cx_Oracle. The documentation for this drive is good and detailed with plenty of examples available on GitHub. Hopefully there isn’t anything new in this post, but it is my experiences and what I did.
1. Install Oracle Client
The Python driver requires Oracle Client software to be installed. Go here, download and install. It’s a straightforward install. Make sure the directories are added to the search path.
2. Download and install cx_Oracle
You can use pip3 to do this.
pip3 install cx_Oracle Collecting cx_Oracle Downloading cx_Oracle-6.1.tar.gz (232kB) 100% |████████████████████████████████| 235kB 679kB/s Building wheels for collected packages: cx-Oracle Running setup.py bdist_wheel for cx-Oracle ... done Stored in directory: /Users/brendan.tierney/Library/Caches/pip/wheels/0d/c4/b5/5a4d976432f3b045c3f019cbf6b5ba202b1cc4a36406c6c453 Successfully built cx-Oracle Installing collected packages: cx-Oracle Successfully installed cx-Oracle-6.1
3. Create a connection in Python
Now we can create a connection. When you see some text enclosed in angled brackets <>, you will need to enter your detailed for your schema and database server.
# import the Oracle Python library import cx_Oracle # define the login details p_username = "" p_password = "" p_host = "" p_service = "" p_port = "1521" # create the connection con = cx_Oracle.connect(user=p_username, password=p_password, dsn=p_host+"/"+p_service+":"+p_port) # an alternative way to create the connection # con = cx_Oracle.connect('/@/:1521') # print some details about the connection and the library print("Database version:", con.version) print("Oracle Python version:", cx_Oracle.version) Database version: 188.8.131.52.0 Oracle Python version: 6.1
4. Query some data and return results to Python
In this example the query returns the list of tables in the schema.
# define a cursor to use with the connection cur = con.cursor() # execute a query returning the results to the cursor cur.execute('select table_name from user_tables') # for each row returned to the cursor, print the record for row in cur: print("Table: ", row) Table: ('DECISION_TREE_MODEL_SETTINGS',) Table: ('INSUR_CUST_LTV_SAMPLE',) Table: ('ODMR_CARS_DATA',)
Now list the Views available in the schema.
# define a second cursor cur2 = con.cursor() # return the list of Views in the schema to the cursor cur2.execute('select view_name from user_views') # display the list of Views for result_name in cur2: print("View: ", result_name) View: ('MINING_DATA_APPLY_V',) View: ('MINING_DATA_BUILD_V',) View: ('MINING_DATA_TEST_V',) View: ('MINING_DATA_TEXT_APPLY_V',) View: ('MINING_DATA_TEXT_BUILD_V',) View: ('MINING_DATA_TEXT_TEST_V',)
5. Query some data and return to a Panda in Python
Pandas are commonly used for storing, structuring and processing data in Python, using a data frame format. The following returns the results from a query and stores the results in a panda.
# in this example the results of a query are loaded into a Panda # load the pandas library import pandas as pd # execute the query and return results into the panda called df df = pd.read_sql_query("SELECT * from INSUR_CUST_LTV_SAMPLE", con) # print the records returned by query and stored in panda print(df.head()) CUSTOMER_ID LAST FIRST STATE REGION SEX PROFESSION \ 0 CU13388 LEIF ARNOLD MI Midwest M PROF-2 1 CU13386 ALVA VERNON OK Midwest M PROF-18 2 CU6607 HECTOR SUMMERS MI Midwest M Veterinarian 3 CU7331 PATRICK GARRETT CA West M PROF-46 4 CU2624 CAITLYN LOVE NY NorthEast F Clerical BUY_INSURANCE AGE HAS_CHILDREN ... MONTHLY_CHECKS_WRITTEN \ 0 No 70 0 ... 0 1 No 24 0 ... 9 2 No 30 1 ... 2 3 No 43 0 ... 4 4 No 27 1 ... 4 MORTGAGE_AMOUNT N_TRANS_ATM N_MORTGAGES N_TRANS_TELLER \ 0 0 3 0 0 1 3000 4 1 1 2 980 4 1 3 3 0 2 0 1 4 5000 4 1 2 CREDIT_CARD_LIMITS N_TRANS_KIOSK N_TRANS_WEB_BANK LTV LTV_BIN 0 2500 1 0 17621.00 MEDIUM 1 2500 1 450 22183.00 HIGH 2 500 1 250 18805.25 MEDIUM 3 800 1 0 22574.75 HIGH 4 3000 2 1500 17217.25 MEDIUM [5 rows x 31 columns]
6. Wrapping it up and closing things
Finally we need to wrap thing up and close our cursors and our connection to the database.
# close the cursors cur2.close() cur.close() # close the connection to the database con.close()
Watch out for more blog posts on using Python with Oracle, Oracle Data Mining and Oracle R Enterprise.