Connecting to Oracle ADB from Python

Connecting to Oracle ADB

Connect to Oracle ATP or ADW from Python

I realised that in some of my previous posts I didn't really detail connecting to ATP and ADW. Here's a slight more in depth walkthough.

Connecting to Oracle Autonomous Transaction Processing or Autonomous Datawarehouse is pretty simple from Python. It requires only a few things

  • Oracle Instant Client (Or alternative)
  • A Python environment.
  • The Oracle_CX Python driver Module.
  • A valid wallet for an ATP or ADW service

Let's go through each of these in turn

Oracle Instant Clent

The next step is pretty straight forward. You can download the oracle instant client from here.

instant client screen shot

You'll only need the basic package. Unzip the downloaded file into a suitable location. It's worth pointing out that on Linux this step is even easier. You can now use yum to install the instant client direct from the command line. You can find details on how to configure it here

Python environment

There's plenty of guides out there that show you how to install python on your windows or mac. If you haven't done this already This guide is a good place to start. I'm assuming that you've also gone through the steps of installing pip. If not you can follow this simple guide. I'd also advice you create a virtual environment with virtualenv before you doing anything else. It's considered best practice and isolates you from current or future library conficts.

First lets create our virtual env

virtualenv adb_virt_env

And then active it (I'm assuming linux or mac)

source adb_virt_env/bin/activate

The next step is to install the Python driver. This is as simple as

pip install cx_Oracle

And thats all we need to do at this stage inside to setup our Python environment.

Oracle ADW or ATP Wallet

The final thing we need is the wallet containing the credential and connect string details to enable us to connect to ATP or ADW. You'll need to log onto Oracle OCI console to do this unless have been provided the wallet by a colleague. Simply navigate to your ATP or ADW instance and follow the instructions below.

download wallet screen shot

While it's not necessary we'll download and unzip the wallet into the virtual directory we've created (adb_virt_env).

$> ls
bin                cx_Oracle-doc      include            lib                pip-selfcheck.json
$> unzip
  inflating: cwallet.sso             
  inflating: tnsnames.ora            
  inflating: truststore.jks          
  inflating: sqlnet.ora              
  inflating: ewallet.p12             
  inflating: keystore.jks
$> ls
bin                cx_Oracle-doc      include            lib                pip-selfcheck.json tnsnames.ora
cwallet.sso        ewallet.p12        keystore.jks   sqlnet.ora         truststore.jks

Next we need to edit the sqlnet.ora file to reflect the location where it's located. Currently for my environment it looks like

WALLET_LOCATION = (SOURCE = (METHOD = file) (METHOD_DATA = (DIRECTORY="?/network/admin")))

We'll need to change the DIRECTORY parameter to our virtual environment. In my case /Users/dgiles/Downloads/adb_virt_env. So for my environment it will look like

WALLET_LOCATION = (SOURCE = (METHOD = file) (METHOD_DATA = (DIRECTORY="/Users/dgiles/Downloads/adb_virt_env")))

We should also take a look at tnsnames.ora to see which services we'll be using. You can do this by taking a look in the tnsnames.ora file. There's likely to by lots of entries if you have lots of ATB or ADW instances in you OCI compartment. In my instance I'll be using a connect string called sbatp_medium which has a medium priority but pick the one appropriate to your environment.

sbatp_medium = (description= (address=(protocol=tcps)(port=1522)(
        ",OU=Oracle BMCS US,O=Oracle Corporation,L=Redwood City,ST=California,C=US"))   )

We'll only need to remember its name for the next step.

The Code

Finally we're ready to write some code. The first step is to import the modules we'll need. In this case it's just cx_oracle and os

In [16]:
import cx_Oracle
import os

We need to set the environment variable TNS_ADMIN to point at our directory (adb_virt_env) where all of the files from our wallet are located.

In [17]:
os.environ['TNS_ADMIN'] = '/Users/dgiles/Downloads/adb_virt_env'

And now we can simply connect to ATP or ADW instance using a standard Python database connect operation using the connect string we remebered from the tnsnames.ora file. NOTE : I'm assuming you've created a user in the database or you're using the admin user created for this instance.

In [18]:
connection = cx_Oracle.connect('admin', 'ReallyLongPassw0rd', 'sbatp_medium')

And thats it... From here on in we can use the connection as it was a local database.

In [19]:
cursor = connection.cursor()
rs = cursor.execute("select 'Hello for ADB' from dual")
[('Hello for ADB',)]

Update to MonitorDB

Just a quick one I've update MonitorDB to enable it to use wallets. So it can now run against Oracle Autonomous Transaction Processing and Oracle Autonomous Data Warehouse.

You can add the location in the configuration file or

Screenshot of IntelliJ IDEA (22-01-2019, 18-48-21)

On the command line

Screenshot of Terminal (22-01-2019, 18-51-09)

I've also compiled it for Java8 and used the latest jdbc drivers.

You can find it here

Oracle SODA Python Driver and Jupyter Lab


Oracle SODA Python Driver and Jupyter Lab

This workbook is divided into two sections the first is a quick guide to setting up Jupyter Lab (Python Notebooks) such that it can connect to a database running inside of OCI, in this case an ATP instance. The second section uses the JSON python driver to connect to the database to run a few tests. This notebook is largely a reminder to myself on how to set this up but hopefully it will be useful to others.

Setting up Python 3.6 and Jupyter Lab on our compute instance

I won't go into much detail on setting up ATP or ADW or creating a IaaS server. I covered that process in quite a lot of detail here. We'll be setting up something similar to the following

OCI Cloud

Once you've created the server You'll need to logon to the server with the details found on the compute instances home screen. You just need to grab it's IP address to enable you to logon over ssh.

OCI Cloud

The next step is to connect over ssh to with a command similar to

ssh opc@
Enter passphrase for key '/Users/dgiles/.ssh/id_rsa': 
Last login: Wed Jan  9 20:48:46 2019 from host10.10.10.1

In the following steps we'll be using python so we need to set up python on the server and configure the needed modules. Our first step is to use yum to install python 3.6 (This is personal preference and you could stick with python 2.7). To do this we first need to enable yum and then install the environment. Run the following commands

sudo yum -y install yum-utils
sudo yum-config-manager --enable ol7_developer_epel
sudo yum install -y python36
python3.6 -m venv myvirtualenv
source myvirtualenv/bin/activate

This will install python and enable a virtual environment for use (our own Python sand pit). You can make sure that python is installed by simply typing python3.6 ie.

$> python3.6
Python 3.6.3 (default, Feb  1 2018, 22:26:31) 
[GCC 4.8.5 20150623 (Red Hat 4.8.5-16)] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> quit()

Make sure you type quit() to leave the REPL environment.

We now need to install the needed modules. You can do this one by one or simply use the following file requirements.txt and run the following command

pip -p requirements.txt

This will install all of the need python modules for the next step which is to start up Jupyter Lab.

Jupyter Lab is an interactive web based application that enables you do interactively run code and document the process at the same time. This blog is written in it and the code below can be run once your environment is set up. Vists the website to see more details.

To start jupyer lab we run the following command.

nohup jupyter-lab --ip= &

Be aware that this will only work if you have activated you virtual environment. In out instance we did this with with the command source myvirtualenv/bin/activate. At this point the jupyter-lab is running in the background and and is listening (by default) on port 8888. You could start a desktop up and use VNC to view the output. However I prefer to redirect the output to my own desktop over ssh. To do this you'll need to run the following ssh command from your desktop

ssh -N -f -L 5555:localhost:8888 opc@

Replacing the IP address above with the one for your compute instance

This will redirect the output of 8888 to port 5555 on your destop. You can then connect to it by simply going to the following url http://localhost:5555. After doing this you should see a screen asking you input a token (you'll only need to do this once). You can find this token inside of the nohup.out file running on the compute instance. It should be near the top of the file and should look something like

[I 20:49:12.237 LabApp]

Just copy the text after "token=" and paste it in to the dialogue box. After completing that step you should see something like this

Jupyter Lab

You can now start creating your own notebooks or load this one found here. I'd visit the website to familiarise yourself on how the notebooks work.

Using Python and the JSON SODA API

This section will walk through using The SODA API with Python from within the Jupyter-lab envionment we set up in the previous section. The SODA API is a simple object API that enables developers persist and retrieve JSON documents held inside of the Oracle Database. SODA drivers are available for Java, C, REST, Node and Python.

You can find the documentation for this API here

To get started we'll need to import the need the following python modules

In [11]:
import cx_Oracle
import keyring
import os
import pandas as pd

We now need to set the location of the directory containing the wallet to enable us to connect to the ATP instance. Once we've done that we can connect to the Oracle ATP instance and get a SODA object to enable us to work with JSON documents. NOTE : I'm using the module keyring to hide the password for my database. You should replace this call with the password for your user.

In [20]:
os.environ['TNS_ADMIN'] = '/home/opc/Wallet'
connection = cx_Oracle.connect('json', keyring.get_password('ATPLondon','json'), 'sbatp_tpurgent')
soda = connection.getSodaDatabase()

We now need to create JSON collection and if needed add any additional indexes which might accelerate data access.

In [21]:
    collection = soda.createCollection("customers_json_soda")
    collection.createIndex({ "name"   : "customer_index",
                          "fields" : [ { "path"     : "name_last",
                          "datatype" : "string"}]})
except cx_Oracle.DatabaseError as ex:
    print("It looks like the index already exists : {}".format(ex))

We can now add data to the collection. Here I'm inserting the document into the database and retrieving it's key. You can find find some examples/test cases on how to use collections here

In [22]:
customer_doc = {"id"          : 1,
       "name_last"    : "Giles",
       "name_first"   : "Dom",
       "description"  : "Gold customer, since 1990",
       "account info" : None,
       "dataplan"     : True,
       "phones"       : [{"type" : "mobile", "number" : 9999965499},
                         {"type" : "home",   "number" : 3248723987}]}
doc = collection.insertOneAndGet(customer_doc)

To fetch documents we could use SQL or Query By Example (QBE) as shown below. You can find further details on QBE here. In this example there should just be a single document. NOTE: I'm simply using pandas DataFrame to render the retrieved data but it does highlight how simple it is to use the framework for additional analysis at a later stage.

In [23]:
documents = collection.find().filter({'name_first': {'$eq': 'Dom'}}).getDocuments()
results = [document.getContent() for document in documents]    
account info dataplan description id name_first name_last phones
0 None True Gold customer, since 1990 1 Dom Giles [{'type': 'mobile', 'number': 9999965499}, {'t...

To update records we can use the replaceOne method.

In [24]:
document = collection.find().filter({'name_first': {'$eq': 'Dom'}}).getOne()
updated = collection.find().key(doc.key).replaceOne({"id"          : 1,
       "name_last"    : "Giles",
       "name_first"   : "Dominic",
       "description"  : "Gold customer, since 1990",
       "account info" : None,
       "dataplan"     : True,
       "phones"       : [{"type" : "mobile", "number" : 9999965499},
                         {"type" : "home",   "number" : 3248723987}]},)

And just to make sure the change happened

In [25]:
data = collection.find().key(document.key).getOne().getContent()
account info dataplan description id name_first name_last phones
0 None True Gold customer, since 1990 1 Dominic Giles [{'type': 'mobile', 'number': 9999965499}, {'t...

And finally we can drop the collection.

In [26]:
except cx_Oracle.DatabaseError as ex:
    print("We're were unable to drop the collection")
In [27]:

Creating, Querying and Loading Data in to the Oracle Autonomous Data Warehouse


Oracle Autonomous Data Warehouse (ADW) access via Python

The following shows how to access the Oracle Autonomous Data Warehouse using Python and how to load the data using the DBMS_CLOUD package via the cx_Oracle module. This is obviously simpler via the Graphical front end or SQL Developer but using Python provdes a simple scriptable model whilst hiding some of the complexities of using native REST APIs.

This simple example assumes that you've got an Oracle Cloud account and that you've created or got access an ADW database. You'll also have to download the credentials file to provide SQL*Net access. You can find the details on how to do that here. We'll be using Python in this short example but most of what we're doing could be achieved using the GUI and/or REST Calls.

Connecting to ADW Instance

To start with we'll make sure we can connect to the ADW Instance we've previously created. To do that we need to import the required libraries. If you dodn't have these I reccommend using PIP (and virtualenv)

In [ ]:
import cx_Oracle
import keyring
import os

We need to use an environment variable to reflect the location of the downloaded credentials files to be used by SQL*Net.

In [ ]:
os.environ['TNS_ADMIN'] = '/Users/dgiles/Downloads/wallet_DOMSDB'

This is equivlent to bash export TNS_ADMIN=/Users/dgiles/Downloads/wallet_DOMSDB and points to the unzipped directory containing the tnsnames.ora, sqlnet.ora etc. NOTE: you'll need to update the sqlnet.ora to ensure the wallet points to the same directory specified in the TNS_ADMIN environment variable. i.e.

WALLET_LOCATION = (SOURCE = (METHOD = file) (METHOD_DATA = (DIRECTORY="/Users/dgiles/Downloads/wallet_DOMSDB")))

In the example above I've changed DIRECTORY to the location where I downloaded and unzipped the credentials file.

The next steps are to connect to the Oracle ADW instance. In the example below I've store my password using the Python Module "keyring". I'm also using the jupyter notebook magic sql functionality. We'll test the connection using the admin user and the connect string "domsdb_medium" which is one of the services included in the tnsnames.ora file.

In [ ]:
%load_ext sql
In [257]:
password = keyring.get_password('adw','admin')
%sql oracle+cx_oracle://admin:$password@domsdb_medium
'Connected: admin@None'
In [258]:
%%sql admin@domsdb_medium
select 1 from dual
0 rows affected.

Generating test data

We've connected to the oracle database and can now start uploading data to the instance. In this example we'll use datagenerator to generate the data into flat files and then place these on Oracle Object Store and load them from there.

The first step is to install datagenerator. You can find details on how to do that here. We can now simply generate data for the "SH" benchmark.

In [259]:
import subprocess

# Change the 2 following parameters to reflect your environment
generated_data_dir = '/Users/dgiles/Downloads/generated_data'
datagenerator_home = '/Users/dgiles/datagenerator'

# Change the following paramters relating to the way datagenerator will create the data
scale = 100
parallel = 8

dg_command = '{dg}/bin/datagenerator -c {dg}/bin/sh.xml -scale {s} -cl -f -d {gdd} -tc {p}'.format(
    dg = datagenerator_home,
    s = scale,
    gdd = generated_data_dir,
    p = parallel
# Typically we'd use a command similiar to the one below but since we're in a notebook it's easier to use the default functionality
# p = subprocess.Popen(command, shell=True, stdout=subprocess.PIPE)
# (output, err) = p.communicate()

/Users/dgiles/datagenerator/bin/datagenerator -c /Users/dgiles/datagenerator/bin/sh.xml -scale 100 -cl -f -d /Users/dgiles/Downloads/generated_data -tc 8
Started Datagenerator, Version

|           Datagenerator Run Stats        |
Connection Time                        0:00:00.000
Data Generation Time                   0:00:26.274
DDL Creation Time                      0:00:00.000
Total Run Time                         0:00:26.284
Rows Inserted per sec                        5,367
Data Generated (MB) per sec                    0.5
Actual Rows Generated                      137,000
Commits Completed                                0
Batch Updates Completed                          0

We should now have a series of files in the "generated_data_dir" directory. These will be a mix of csv files, create table scripts, loader scripts etc.

In [255]:
!ls {generated_data_dir}
CHANNELS.csv                   SALES.csv
CHANNELS.ctl                   SALES.ctl
CUSTOMERS.csv                  constraints.sql
CUSTOMERS.ctl                  createindexes.sql
PRODUCTS.csv                   createsequences.sql
PRODUCTS.ctl                   createtables.sql
PROMOTIONS.csv                 droptables.sql

Uploading the data to the Oracle Object Store

We're really only interested in the "csv" files so we'll upload just those. But before we do this we'll need to establish a connection to the Oracle Object Store. I give some detail behind how to do this in this notebook. I'll be using the object storage out of the Frankfurt Region.

In [256]:
import oci
import ast

my_config = ast.literal_eval(keyring.get_password('oci_opj','doms'))
my_config['region'] = 'eu-frankfurt-1'
object_storage_client = oci.object_storage.ObjectStorageClient(my_config)
namespace = object_storage_client.get_namespace().data

We've now got a handle to the Oracle Object Store Client so we can now create a bucket which we'll call and upload the "CSV" Files.

In [ ]:
import os, io

bucket_name = 'Sales_Data'

files_to_process = [file for file in os.listdir(generated_data_dir) if file.endswith('csv')]

    create_bucket_response = object_storage_client.create_bucket(
except Exception as e:

for upload_file in files_to_process:
    print('Uploading file {}'.format(upload_file))
    object_storage_client.put_object(namespace, bucket_name, upload_file,,upload_file),'r'))    

We need to create an authentication token that can be used by the ADW instance to access our Object storage. To do this we need to create an identity client.

In [ ]:
indentity_client = oci.identity.IdentityClient(my_config)
In [ ]:
token = indentity_client.create_auth_token(
        description = "Token used to provide access to newly loaded files"
    user_id = my_config['user']

Creating users and tables in the ADW Instance

The following steps will feel very familiar to any DBA/developer of and Oracle database. We need to create a schema and assocated tables to load the data into.

First we'll need to create a user/schema and grant it the appropriate roles

In [ ]:
%sql create user mysh identified by ReallyLongPassw0rd default tablespace data

Grant the "mysh" user the DWROLE

In [ ]:
%sql grant DWROLE to mysh
In [ ]:
%sql oracle+cx_oracle://mysh:ReallyLongPassw0rd@domsdb_medium

We can now create the tables we'll use to load the data into.

In [ ]:
%%sql mysh@domsdb_medium
  COUNTRY_ID           NUMBER       NOT NULL,
In [ ]:
%%sql mysh@domsdb_medium
  TIME_ID       DATE       NOT NULL,
In [ ]:
%%sql mysh@domsdb_medium
  CUST_ID                 NUMBER NOT NULL,
  EDUCATION               VARCHAR2(21),
  OCCUPATION              VARCHAR2(21),
  AFFINITY_CARD           NUMBER(10),
  Y_BOX_GAMES             NUMBER(10),
  COMMENTS                VARCHAR2(4000)
In [ ]:
%%sql mysh@domsdb_medium
  CUST_ID                NUMBER       NOT NULL,
  CUST_GENDER            CHAR(1)      NOT NULL,
  CUST_CITY              VARCHAR2(30) NOT NULL,
  CUST_CITY_ID           NUMBER       NOT NULL,
  COUNTRY_ID             NUMBER       NOT NULL,
  CUST_EMAIL             VARCHAR2(40),
  CUST_TOTAL             VARCHAR2(14) NOT NULL,
  CUST_SRC_ID            NUMBER,
  CUST_EFF_FROM          DATE,
  CUST_EFF_TO            DATE,
  CUST_VALID             VARCHAR2(1)
In [ ]:
%%sql mysh@domsdb_medium
In [ ]:
%%sql mysh@domsdb_medium
  PROD_ID               NUMBER(6)      NOT NULL,
  PROD_NAME             VARCHAR2(50)   NOT NULL,
  PROD_DESC             VARCHAR2(4000) NOT NULL,
  SUPPLIER_ID           NUMBER(6)      NOT NULL,
  PROD_STATUS           VARCHAR2(20)   NOT NULL,
  PROD_TOTAL            VARCHAR2(13)   NOT NULL,
  PROD_TOTAL_ID         NUMBER         NOT NULL,
  PROD_SRC_ID           NUMBER,
  PROD_EFF_TO           DATE,
  PROD_VALID            VARCHAR2(1)
In [ ]:
%%sql mysh@domsdb_medium
  PROMO_ID             NUMBER(6)    NOT NULL,
  PROMO_COST           NUMBER(10)   NOT NULL,
In [245]:
%%sql mysh@domsdb_medium
SELECT udate time_id,
  TO_CHAR(udate,'Day'Winking day_name,
  TO_CHAR(udate,'DD'Winking day_number_in_month,
  TO_CHAR(udate,'DDD'Winking day_number_in_year,
  TO_CHAR(udate,'YYYY' ) calendar_year,
  TO_CHAR(udate,'Q' ) calendar_quarter_number,
  TO_CHAR(udate,'MM' ) calendar_month_number,
  TO_CHAR(udate,'WW' ) calendar_week_number,
  TO_CHAR(udate,'YYYY-MM'Winking calendar_month_desc,
  TO_CHAR(udate,'YYYY-Q'Winking calendar_quarter_desc
  (SELECT to_date('31/12/1994','DD/MM/YYYY'Winking+rownum udate
  FROM all_objects
  WHERE to_date('31/12/1994','DD/MM/YYYY'Winking+rownum <= to_date( '31/12/2014','DD/MM/YYYY'Winking
7305 rows affected.
In [248]:
%%sql mysh@domsdb_medium
select * from tab
0 rows affected.
tname tabtype clusterid

Copying the data from the object store

We need to add the authorisation token to the newly created schema to allow it to access the object stores files. We can't do this using the sql magic syntax we've been using till this point so we'll do it using standard cx_Oracle calls.

In [ ]:
connection = cx_Oracle.connect('mysh', 'ReallyLongPassw0rd', 'domsdb_medium')
cursor = connection.cursor();
In [ ]:
cursor.callproc('DBMS_CLOUD.create_credential', keywordParameters = {'credential_name':'SALES_DATA_AUTH',

We can access the object storage using a url of the the format

https://swiftobjectstorage.<region><tenancy>/<bucket name>/<object name>

We can use this to dynamically generate a url for each of the objects inside of the bucket we've just created and use the DBMS_CLOUD package to copy the data into the ADW instance. The code below gets all of the names of the tables we've just created and the loops through each table copying the associated csv file into the ADW instance.

In [ ]:
from tqdm import tqdm

format = '''{"delimiter" : ",",
          "skipheaders" : 1,
          "ignoremissingcolumns" : "true",
          "removequotes" : "true",
          "dateformat" : "DD-MON-YYYY HH24:MI:SS",
          "blankasnull" : "true"}'''

file_location = '''https://swiftobjectstorage.{region}{tenancy}/{bucket_name}/{table_name}.csv'''
region = my_config['region']
tenancy= 'oracleonpremjava'

rs = cursor.execute("select table_name from user_tables where table_name not like 'COPY%'")
rows = rs.fetchall()
for row in tqdm(rows):
    url = file_location.format(region=region, tenancy=tenancy, bucket_name=bucket_name, table_name=row[0])
    cursor.callproc('DBMS_CLOUD.copy_data', keywordParameters= {'table_name':row[0],
                                                           'format': format

We can now take a look and see how many rows we've loaded into the tables

In [261]:
rs = cursor.execute("select table_name from user_tables where table_name not like 'COPY%'")
rows = rs.fetchall()

for row in rows:
    rs2 = cursor.execute("select count(*) from {}".format(row[0]))
    rows2 = rs2.fetchone()
    print('{tn: <35}{rc:>10,}'.format(tn=row[0],rc=rows2[0]))
COUNTRIES                                  22
SALES                                  97,799
CUSTOMERS                              19,599
CHANNELS                                    4
PRODUCTS                                   71
PROMOTIONS                                502
TIMES                                   7,305

It's now possible to run standard queries against the newly loaded data. No need to create anything else (indexes etc.)

In [249]:
%%sql mysh@domsdb_medium
SELECT channels.channel_desc, 
       TO_CHAR(SUM(amount_sold), '9,999,999,999'Winking SALES$
FROM sales, customers, times, channels, countries
WHERE sales.time_id=times.time_id 
  AND sales.cust_id=customers.cust_id 
  AND customers.country_id = countries.country_id
  AND sales.channel_id = channels.channel_id 
  AND channels.channel_desc IN ('Tele Sales','Internet'Winking 
  AND times.calendar_year = '2006' 
  AND countries.country_iso_code IN ('GB','DE','FR','DK'Winking
0 rows affected.
channel_desc country_iso_code sales$
Internet DE          3,130
Internet DK          1,676
Internet FR          2,545
Internet GB          2,320
Internet None          9,671
Tele Sales DE          3,077
Tele Sales DK          3,116
Tele Sales FR          3,184
Tele Sales GB          2,386
Tele Sales None         11,763
None None         21,434

Tidying up the object store and database

You can run the following steps if you want to remove all of the tables from the schema and purge the object store of files. Lets start by removing the tables. NOTE : The code below will remove all of the tables from the schema. Make sure you've not got anything in the schema that you want to keep before running it.

In [ ]:
from tqdm import tqdm

rs = cursor.execute("select table_name from user_tables")
rows = rs.fetchall()

for row in tqdm(rows):
    rs2 = cursor.execute("drop table {} purge".format(row[0]))

And then removing the object store files and bucket

In [ ]:
object_list = object_storage_client.list_objects(namespace, bucket_name)

for o in
    print('Deleting object {}'.format(
    object_storage_client.delete_object(namespace, bucket_name,

print('Deleting bucket')    
response = object_storage_client.delete_bucket(namespace, bucket_name)

Making the alert log just a little more readable

One of the most valuable sources of information about what the Oracle database has done and is currently doing is the alert log. It's something that every Oracle Database professional should be familiar with. So what can you do to improve you chances of not missing important pieces of info? The obvious answer is that you should use a tool like Enterprise Manager. This is particularly true if you are looking after hundreds of databases.

But what if you are only looking after one or two or just testing something out? Well the most common solution is to simply tail the alert log file.

The only issue is that it's not the most exciting thing to view, this of course could be said for any terminal based text file. But there are things you can do to make it easier to parse visually and improve your chances of catching an unexpected issue.

The approach I take is to push the alert log file through python and use the various libraries to brighten it up. It's very easy to go from this (tail -f)

Screenshot of ScreenFloat (20-03-2018, 08-25-26)

To this

Screenshot of ScreenFloat (20-03-2018, 08-25-58)

The reason this works is that python provides a rich set of libraries which can add a little bit of colour and formatting to the alert file.

You can find the code to achieve this in the gist below

Just a quick note on installing this. You'll need either python 2.7 or 3 available on your server.

I'd also recommend installing pip and then the following libraries

pip install humanize psutil colorama python-dateutil

After you've done that it's just a case of running the script. If you have $ORACLE_BASE and $ORACLE_SID set the library will try and make a guess at the location of the alert file. i.e


But if that doesn't work or you get an error you can also explicitly specify the location of the alert log with something like

python -a $ORACLE_BASE/diag/rdbms/orcl/orcl/trace/alert_orcl.log

This script isn't supposed to be an end product just a simple example of what can be achieved to make things a little easier. And whilst I print information like CPU load and Memory there's nothing to stop you from modifying the script to display the number of warnings or errors found in the alert log and update it things change. Or if you really want to go wild implement something similar but a lot more sophisticated using python and curses

The age of "Terminal" is far from over….

Changing the size of redo logs in python

I create a lot of small databases to do testing on. The trouble is that I often need to change the size of redo log files when I'm testing large transaction workloads or loading a lot of data. Now there are lots of better ways to do whats shown in the code below but this approach gave me the chance to keep brushing up my python skills and use the might cx_oracle driver. The following should never be considered anything but a nasty hack but it does save me a little bit of time i.e. don't use this on anything but a test database… Clearly the sensible way to do this is to write my own scripts to build databases.

The following code works it's way through the redo log files drops one thats inactive and then simply recreates it. It finished when it's set all of the redo to the right size.

Running the script is simply a case of running it with the parameters shown below

python ChangeRedoSize -u sys -p welcome1 -cs myserver/orclcdb --size 300

Note : the user is the sysdba of the container database if you are using the multitenant arhcitecture and the size is in Mega Bytes.

You should then see something similar to the following

Current Redo Log configuration
| Group No. | Thread No. | Sequence No. | Size (MB) | No of Members |  Status  |
|     1     |     1      |     446      | 524288000 |       1       | INACTIVE |
|     2     |     1      |     448      | 524288000 |       1       | CURRENT  |
|     3     |     1      |     447      | 524288000 |       1       |  ACTIVE  |
alter system switch logfile
alter system switch logfile
alter database drop logfile group 2
alter database add logfile group 2 size 314572800
alter system switch logfile
alter database drop logfile group 1
alter database add logfile group 1 size 314572800
alter system switch logfile
alter system switch logfile
alter system switch logfile
alter system switch logfile
alter database drop logfile group 3
alter database add logfile group 3 size 314572800
alter system switch logfile
All logs correctly sized. Finishing...
New Redo Log configuration
| Group No. | Thread No. | Sequence No. | Size (MB) | No of Members |  Status  |
|     1     |     1      |     455      | 314572800 |       1       |  ACTIVE  |
|     2     |     1      |     454      | 314572800 |       1       | INACTIVE |
|     3     |     1      |     456      | 314572800 |       1       | CURRENT  |

Notes on pre-parsing data for Oracle data loads

Sometimes data simply isn't in a form that is easy to load into an Oracle database i.e. column form. It would be great if everybody exchanged data in a simple CSV form with a single file to table mapping. Sadly that isn't the case and sometimes you have to do a little work to get it into a form thats useable. A recent benchmark highlighted this issue very well. The customer provided the data in compressed CSV form (so far so good) but the data was held in key value pairs (not so good). They also provided us with a mapping file that describes how it all fits together. 

Now typically the approach many people would take would be to develop some form of program that parses all of the data and writes it to staging  area and then loads all of it in one go to the target database. I make no criticism of this approach since it works well and as long as its not time critical. It's by far the simplest method. However Im a big fan of taking advantage of whats already available and one of the most underused and powerful features of the Oracle database is  the preparser. It enables you to pipeline various operations so they all run as quickly as possible. So going back to my benchmark we used this approach to load data into out target database. It consisted of 4 steps
  • Read the data of the filesystem as efficiently as possible and write it to stdout
  • Read from stdin and Unzip the the data writing it to stdout
  • Read from stdin into a java program to do the key value mapping and error detection/correction writing the output to stdout
  • Read from stdin into sqlloader
I will at this time point out I'm not really using Oracle's pre-parser I'm just using good old "Pipes" but why this is important will become clearer later This approach gave us a great deal of flexibility and simplified the code we had to write. It operates in some respects as a serialised map reduce flow but I'll come back to that another day and explain how it can be integrated directly into a massively parallel approach. It's also possible to get Java to natively read the zipped file as well having said that I offloaded that process to the os to enable me to use different compression formats when needed.

The java program simple reads from stdin and writes to stdout. To handle key value pairs just required the program to read the mapping file in and split and parse the values from stdin. The data was then written to stdout in a well know order.

Java extract from my program... 

            BufferedReader br = new BufferedReader(new InputStreamReader(, OneMB);
            String line = null;
            HashMap keyValuePairs = null;
            MyTokenizer mt = null;
            while ((line = br.readLine()) != null) {
                keyValuePairs = new HashMap(200);
                mt = new MyTokenizer(line, delimitor);
                for (String token : mt) {
                    int loc = token.indexOf("=");
                    if (loc != -1) {
                        String i = token.substring(0, loc);
                        String s = token.substring(loc + 1, token.length());
                        keyValuePairs.put(i, s);
                StringBuffer outRec = new StringBuffer(1000);
                outRec.append(checkForNull(keyValuePairs.get("uniqueID"), "")).append(seperator);
                // mapping logic similar to above repeats

All that was needed for sqlloader to process the files was a control file that understood the order of the columns and any additional formatting. 

One of the additional benefits is that we can load the data via "direct path" and implement other features such as multi table insert. The Java preparser enables you to add all of the additional formatting to make this a trivial process.

The following diagram illustrates the process.

This equates into a Unix/Linux statement such as

/bin/dd if=myverybigfile.txt bs=1024k status=noxfer 2>/dev/null | /bin/gunzip -c | java -classpath /home/oracle/loader.jar com.dom.KeyValueParserStdIn | sqlloader bench/bench control=kv.ctl data=\"-\" direct=TRUE;

NOTE : one thing you may have noticed is that Im using dd to do 1MB I/Os. This just an efficiency operation and works well on structures such as DBFS, you could skip this part of the operation if needed.

Which brings us onto external tables and the preparser

External Tables and pre-parsers 

As I mentioned earlier I like to take advantage of functionality that's already available and one of those features in the Oracle database is external tables. I don't intend to go into much detail as to why you should use external tables other than they do much of the heavy lifting for you and they provide a seamless interface between the filesystem and the database. They effectively make files look like tables. 

This means it's trivial to implement parallelism for our pre-parser. We don't need to worry about how to handle the files and how to schedule everything, external tables take care of all of that for you. In our benchmark we used them in the following way

Our previous pipeline remains the same except that we don't need sqlloader its all managed by the table definition itself. So we end up with something similar to the following for the table definition (I've abbreviated it quite substantially and highlighted the important bits)

create table staging_ext_mydata_jan01
(    uniqueid NUMBER,
-- Lots of columns
   TYPE oracle_loader
   DEFAULT DIRECTORY load_dir_jan01
      PREPROCESSOR exec_dir:''
      BADFILE log_dir: 'external.bad'
      LOGFILE log_dir: 'external.log'
fields terminated by '|'
   ( uniqueid char(100),
-- Lots of defintions

One of the things to note is that I've included the pipelined preprocessor inside of a shell script which looks like this 

/bin/dd if=$1 bs=1024k status=noxfer 2>/dev/null | /bin/gunzip -c | java -classpath /home/oracle/loader.jar com.dom.KeyValueParserStdIn

The important part of this script is the parameter ($1) that is passed to the shell script. This is the file name that the external table wants to process. 

A simple select statement from my "staging_ext_mydata_jan01" unzips and parses the data converting it to usable columns. Whats more if I issue the statement in parallel Oracle takes care of creating the processes for me and making sure everything is scheduled in an orderly fashion.

To finish the load we simply used a multi table insert to put the data into the correct tables in an efficient fashion. Using this approach we were able to read zipped files, parse them and insert them into our three target tables at over 1.5 million source records/sec.

Timing groups of SQL operations

Some times I feel like I’ve missed out on a whole chunk on functionality in Oracle products. One little nugget is the “timing” function in SQL*Plus. This allows you to time groups of operations.

Obviously turning on is achieved with the “set timing on” operation. i.e

SQL > set timing on

SQL > select count(1) from all_objects;


Elapsed: 00:00:03.95


Which is great but what if want to time mulitiple operations. Use the timing function and simply give the timer a name, in this case statement timer.

SQL> timing start statement_timer
SQL> select count(1) from all_objects;


SYS@orcl > /


SQL> timing show statement_timer;
timing for: statement_timer
Elapsed: 00:00:30.85

Which times anything that went on in between the timer starting and finishing. In this case also my typing of the commands. Its a fantastic utility for timing stages in a batch job including call outs to os operations.

Datawarehousing benchmark

Recently some of my colleagues and myself worked on a data warehousing benchmark. This comprised of a 1TB data set consisting of a 14 billion row fact table with 5 dimensions of a much smaller size. Nothing outrageous a classical star schema... bread and butter to Oracle. We used the following hardware configuration

2 Intel based servers.
  • 4 CPUs with Dual Cores (hyper threading enabled).
  • 64 GB of memory in each
  • 2 dual ported 4Gb HBAs
  • 2 dual ported 1Gb Nics
The storage consisted of
  • 10 low cost storage arrays
  • 20 controllers
  • 20 trays of disk with 14 15k 36GB drives in each
The fabric consisted of
  • 2 mid range 16 port fibre switches
The network consisted of
  • 1 mid range 16 port 1Gb switch (I know we really should have had two of these)

The point here is that the hardware was low end but with the ability to deliver plenty of CPU and I/O at very attractive price point. For those that aren't that interested in any more of the details the headline figures where that we performed full table scans at over 1.6GB/sec but due to compression we achieved logically three times this figure.

We used a stock 2.6 64 bit Linux kernel we modified the following kernel settings
  • wem, rmem were to set to 512k to reflect the fact that we were going to use 32k block sizes
  • Huge pages were enabled... its a much more sensible way of managing the shared memory required for the SGA
  • We enabled jumbo frames in the switch and set the MTU on the interconnect to 9000bytes
  • We used the deadline scheduler to improve I/O prioritisation.
  • Multipath was used to provide DMP over the 4 paths to disk from each of the servers.

The Oracle database was largely unchanged from the defaults with the exception of the following settings
  • 16GB for SGA and PGA
  • The fact table was compressed
  • 32k block sized to improve compression ratios
  • Two parallel instance groups to ensure execution of certain queries on a given node.
  • Query rewrite was enabled.
  • ocfs2 was used for the quorum and ocr files
  • ASM was used to provide storage for the datafiles etc. Each physical tray of disk was presented to ASM as logical disk of roughly 500Gb.
  • ASMlib was used to persist disk identity and simplify configuration.

The tests consisted of a series of SQL statements that were run serial, concurrently and as a series of streams. I cant give much in the way of specifics for obvious reasons but the machine ran at about 80% utilization during the test most of the queries returned sub minute when run against the full data and many sub second when materialized views were exploited.

However the real benefit came from the fact that low the cost commodity hardware made the system relatively cheap and easy to put together. Like any system there were things we would have liked to have done differently if we'd had the chance... The benchmark was done in 10 days (OS Install, disk, layout, Oracle install, data load etc.) and so things were done in a rush and with a little more thought could have been done much more efficiently... We are fairly certain we could have improved the I/O performance but we didn't have the luxury of experimentation and so we made a call and lived with the consequences... I strongly recommend that any one building one of these systems spends some time with Oracle Orion disk benchmarking system to determine an optimal layout. A recent customer evaluating several possible disk layout configurations showed over 100% difference between two subtly different versions.

On the subject of I/O

One the things that constantly surprises me when talking with clients about hardware for a new database server is that I/O is always at the bottom of the list. Typically the list will look something like this (listed in order of perceived importance)
  • CPUs, have we enough. Fast as possible.

  • Memory, as much as we can put in the box. Oracle don't charge us for that

  • SAN, big as possible.

At this stage the purchase order is usually given the nod and the hardware supplier will ship yet another run of the mill box. Don't get me wrong. Many experienced DBAs have been through this process many times before and realise that not only is the list in the wrong order but its missing some critical components.
  • HBAs, need to specify these in proportion to the CPUs and attached storage

  • NICs, might need a lot of these i.e public, cluster interconnect, storage, management, backup. And typically in multiples for resillience or performance.

  • Backup, are we using the existing backup infrastructure?

I don't blame anyone for this way of thinking, its the way its always been. When discussing a new server the first question that people tend to ask is "So whats this monster packing? 16 CPUs!!!" followed by lots of very macho grunts and hollering. The standard licensing model (not just Oracles) doesn't help. It starts with premise of a CPU describing the power of a server, and to a large degree it does but misses the point of what a database is all about and that's information. Typically that information is held in ones and zeros on a bunch of spinning scrap metal. The real power of a database comes from its ability to aggregate, analyze and process those ones and zeros, turn it into information and push results out to interested parties. Paraphrasing a little "Its all about I/O stupid".

With this in mind I'm constantly surprised by the imbalance of I/O put into servers both disk and network. Its not unusual to see a 4 cpu server running with the latest generation Intel and AMD CPUs but with a single HBA and dual ported NIC. Whilst memory is cheap many of these servers still run 32 bit kernels. This typically means only a small proportion of the database is cached in memory be it in the SGA or file cache (don't me started on file cache). I'd make a rash guess that whilst the size of the memory in a typical database server has increased the average size of the SGA hasn't increased in line with this trend. To make matters worse the typical size of a database has got significantly bigger. This has to lead us to the conclusion that less of the database is cached and as a result a bigger proportion of its is located on disk. As I said this is just a guess but its backed up with real customer engagements. What would be of interest is to have performed an analysis over the last 10 years to see if the wait event for scattered and sequential reads had decreased or increased as a proportion of the total wait event in production databases.

What I'm driving at is the need to move I/O way up the agenda when sizing a server for databases. The number of CPUs needs to be married to the number of I/O channels available. It makes no sense to buy database licenses for a machine that will simply sit and wait on I/O, Its simply wasting money. Equally it makes no sense to stuff a 4 cpu machine full of HBAs for a database application that will perform index lookups on a index that fits comfortably in the cache. Adding HBAs later to an existing server isn't necessarily a simple option either especially for a mission critical application or one that has hard coded paths to disk.

The next obvious question is "well thats well and good but how do I size the ratio of HBAs to CPUs." and in a typically vague fashion I reply "well that depends". The type of application and the type of processor should heavily influence the decision. Certainly the CPU has been winning the race in terms of performance over the last few years and it needs a lot more I/O to keep it busy. But the equation also needs to be balanced with the amount of memory available on the box. A large SGA will certainly reduce the need to visit disk. The best advice I can give is to speak to your hardware supplier and find out what the current state of play is. Also check the latest TPC-C and TPC-H figures show. Whilst these are generally edging towards the extremes of performance it does show what a hardware supplier believed was needed to show their hardware in the best light.


SQL Developer goes production.

SQL Developer has gone production. Congratulations to the entire development team. I've been using it every day now and it feels solid and performs well and I continue to find out new tricks and features each day. If you've not given it ago, try it...


If you haven't tried this tool out I strongly recomend you pop over here. Its a massive step forward for database developers/DBA who really have felt a little neglected by Oracle over the last few years. I've had countless complaints about how SQLPlus/vi/notepad are still used by many as their development tools of choice and how it really isn't good enough. Well I've know about th tool for a while now and have had to keep quite but Im glad the cats out of the bag and its got such positive reviews... especially because of its price... free.

It features much of the functionality you'd expect in a top end development tool plus features that many of its competitors charge top dollar for. The best piece of news is that its an extensible framework and plugins have started to pop up all of the place... One of my particular favourites is at fourthelephant. Perhaps a little over the top for a text man like myself but I appreciate the work that must have gone into it.

They've inspired me to think about putting one together myself... The API is pretty simple and so it shouldn't be too taxing.... Any ideas? drop me a line and I'll see what I can do.

rlwrap : Command line editing in sqlplus

I imagine alot of people in the Linux comunity already use this. However if you new to Linux and struggling with recalling text on the commandline from within sqlplus, rlwrap might be the tool for you... its just a wrapper you put around a command line program and provides command line editing.... I've been using it for the last few months and couldn't live without it

You can find the tool here

Server side failover

I've had a number of emails recently with requests for help with the new server side failover functionaility in 10g release 2. This functionality is described in the Oracle10g release 2 documentation but I've been told its not really obvious.

Let start by explaining what it is. Server side failover allows the sysadmin/dba to configure the profile of connection availablity on the server using a service. Users are effectively unaware of what will happen in the advent of a node in the cluster failing. Previously in (9i and 10g) users needed an entry in their local tnsnames file to describe which nodes they could failover to and which nodes were used to load balance connections on. Unless you used a remote naming service to maintain the connection information every time you added or removed nodes from a cluster it meant an update to potentially hundreds or thousands of tnsnames files.

This was simplified with easy connect in 10g release 1 which allowed the creation and connection to a service specified on the server. For the first time users only needed to connect to a nominated "listener node" and know the name of the service, for example imagine we have a nominated server inside of our organisation called "oracleservice", this of course is the name used for our virtual ip that will float between our cluster of listeners. In 10g release 1 we could create a service called "orderentry" using either dbca or srvctl that would allow our users to connect to it using a connect string of the form

sqlplus soe/soe@//oracleservice/orderentry

This greatly simplifies Oracle network maintenance. In some cases it could mean the removal of tnsnames files from the client or application server. It has other advantages for the DBA as well. If some business event occurs that requires the provisioning of a new application or a new resource profile for a short period of time the DBA can provision it in seconds and trivially remove it when it is no longer required.

Sadly in Oracle10g release 1 this functionality didn't support Transparent Application Failover (TAF), this meant that DBAs still needed to maintain tnsnames files contain a description of what nodes a service could failover on. The good news is that in Oracle10g release 2 this all changed. DBA's could set up a service specifying TAF and the Oracle OCI layer would use this definition provided by the server to describe the load balancing and failover profile.

Implementing this functionality is pretty trivial but there is a step that might catch you out. So lets go through it step by step

To set up the service you can use either Oracle DBCA, the DBMS_SERVICE package, Enterprise Manager or srvctl. The choice is entirely dependent on what you have running. DBCA or Enterprise manager provide the simplest mechanism but you will still have to run the final step using the dbms_service package to tell the database about its failover profile.

I'll use the DBMS_SERVICE package and srvctl for the sake of brevity. In the following example I have a database called db10g2 with two instances db10g21 and db10g22. Im going to create a service called "orderentry" that will provide transparent application failover between the two instances.

The first step is to create the service using srvctl

srvctl add service -d db10g2 -s orderentry -r "db10g21,db10g22" -a "db10g21,db10g22" -P BASIC

and check on its status

$ > srvctl status service -d db10g2 -s "orderentry"
Service orderentry is not running.

So we'll have to start the service first

$ > srvctl start service -d db10g2 -s "orderentry"

if we now use sqlplus connecting as system/sys we can see the service.

SYSTEM@db10g21 > select SERVICE_ID, NAME, NETWORK_NAME, failover_method from dba_services;

id Name Network Name Failover
--- ------------------ ------------------------- ------------
3 orderentry orderentry

The thing to note is that the service hasn't got a failover profile associated with it. So we'll have to modify it using the DBMS_SERVICE package

SYS@db10g21 > get t1.sql
1 begin
3 service_name => 'orderentry',
6 failover_retries => 180,
7 failover_delay => 5);
8* end;
SYS@db10g21 >

if we now select the service information again

id Name Network Name Failover
--- ------------------ ------------------------- ------------
3 orderentry orderentry BASIC

We can now test the service using sqlplus.

sqlplus soe/soe@//node1/orderentry

SQL*Plus: Release - Production on Wed Jan 18 14:11:47 2006

Copyright (c) 1982, 2005, Oracle. All rights reserved.

Connected to:
Oracle Database 10g Enterprise Edition Release - Production
With the Partitioning, Real Application Clusters, OLAP and Data Mining options

SOE@//node1/orderentry >

So all we need to do now is to fire up swingbench and use the service we've created.

[oracle@node1 bin]$ ./charbench -cs //node1/orderentry -dt oci -uc 30 -a
Author : Dominic Giles
Version : 2.2

Results will be written to results.xml.
Users : 30 TPM : 272 Nested TPM : 0

If we log onto the database we can see that the connections have being balanced across the two nodes

SYS@db10g21 >;

1 select instance_name, count(1) usercount, nvl(username,'INTERNAL') user_name,
2 failover_type, failover_method
3 from gv$session s, gv$instance i
4 where s.inst_id = i.inst_id
5 group by instance_name, username, failover_type, failover_method
6* order by username, instance_name
SYS@db10g21 > /

Instance No. of Users Username Fail Over Type Fail Over Method
---------- ------------ ---------- ------------------ ------------------
db10g21 6 SYS NONE NONE
db10g22 6 SYS NONE NONE

so lets shut down of the instances

SQL*Plus: Release - Production on Wed Jan 18 15:28:22 2006

Copyright (c) 1982, 2005, Oracle. All rights reserved.

Connected to:
Oracle Database 10g Enterprise Edition Release - Production
With the Partitioning, Real Application Clusters, OLAP and Data Mining options

SYS@db10g22 > shutdown abort;
ORACLE instance shut down.
SYS@db10g22 >

And re-query the session profile

Instance No. of Users Username Fail Over Type Fail Over Method
---------- ------------ ---------- ------------------ ------------------

There's a lot more thats possible using the service approach to database connection but I'll discuss that in another blog.