OLTP vs Data Warehousing

OLTP Versus Data Warehousing

I’ve tried to explain the difference between OLTP systems and a Data Warehouse to my managers many times, as I’ve worked at a hospital as a Data Warehouse Manager/data analyst for many years. Why was the list that came from the operational applications different than the one that came from the Data Warehouse? Why couldn’t I just get a list of patients that were laying in the hospital right now from the Data Warehouse? So I explained, and explained again, and explained to another manager, and another. You get the picture.
In this article I will explain this very same thing to you. So you know  how to explain this to your manager. Or, if you are a manager, you might understand what your data analyst can and cannot give you.

OLTP

OLTP stands for OLine Transactional Processing. With other words: getting your data directly from the operational systems to make reports. An operational system is a system that is used for the day to day processes.
For example: When a patient checks in, his or her information gets entered into a Patient Information System. The doctor put scheduled tests, a diagnoses and a treatment plan in there as well. Doctors, nurses and other people working with patients use this system on a daily basis to enter and get detailed information on their patients.
The way the data is stored within operational systems is so the data can be used efficiently by the people working directly on the product, or with the patient in this case.

Data Warehousing

A Data Warehouse is a big database that fills itself with data from operational systems. It is used solely for reporting and analytical purposes. No one uses this data for day to day operations. The beauty of a Data Warehouse is, among others, that you can combine the data from the different operational systems. You can actually combine the number of patients in a department with the number of nurses for example. You can see how far a doctor is behind schedule and find the cause of that by looking at the patients. Does he run late with elderly patients? Is there a particular diagnoses that takes more time? Or does he just oversleep a lot? You can use this information to look at the past, see trends, so you can plan for the future.

The difference between OLTP and Data Warehousing

This is how a Data Warehouse works:

The data gets entered into the operational systems. Then the ETL processes Extract this data from these systems, Transforms the data so it will fit neatly into the Data Warehouse, and then Loads it into the Data Warehouse. After that reports are formed with a reporting tool, from the data that lies in the Data Warehouse.

This is how OLTP works:

Reports are directly made from the data inside the database of the operational systems. Some operational systems come with their own reporting tool, but you can always use a standalone reporting tool to make reports form the operational databases.

Pro’s and Con’s

Data Warehousing

Pro’s:

  • There is no strain on the operational systems during business hours
    • As you can schedule the ETL processes to run during the hours the least amount of people are using the operational system, you won’t disturb the operational processes. And when you need to run a large query, the operational systems won’t be affected, as you are working directly on the Data Warehouse database.
  • Data from different systems can be combined
    • It is possible to combine finance and productivity data for example. As the ETL process transforms the data so it can be combined.
  • Data is optimized for making queries and reports
    • You use different data in reports than you use on a day to day base. A Data Warehouse is built for this. For instance: most Data Warehouses have a separate date table where the weekday, day, month and year is saved. You can make a query to derive the weekday from a date, but that takes processing time. By using a separate table like this you’ll save time and decrease the strain on the database.
  • Data is saved longer than in the source systems
    • The source systems need to have their old records deleted when they are no longer used in the day to day operations. So they get deleted to gain performance.

Con’s:

  • You always look at the past
    • A Data Warehouse is updated once a night, or even just once a week. That means that you never have the latest data. Staying with the hospital example: you never knew how many patients are in the hospital are right now. Or what surgeon didn’t show up on time this morning.
  • You don’t have all the data
    • A Data Warehouse is built for discovering trends, showing the big picture. The little details, the ones not used in trends, get discarded during the ETL process.
  • Data isn’t the same as the data in the source systems
    • Because the data is older than those of the source systems it will always be a little different. But also because of the Transformation step in the ETL process, data will be a little different. It doesn’t mean one or the other is wrong. It’s just a different way of looking at the data. For example: the Data Warehouse at the hospital excluded all transactions that were marked as cancelled. If you try to get the same reports from both systems, and don’t exclude the cancelled transactions in the source system, you’ll get different results.

online transactional processing (OLTP)

Pro’s

  • You get real time data
    • If someone is entering a new record now, you’ll see it right away in your report. No delays.
  • You’ve got all the details
    • You have access to all the details that the employees have entered into the system. No grouping, no skipping records, just all the raw data that’s available.

Con’s

  • You are putting strain on an application during business hours.
    • When you are making a large query, you can take processing space that would otherwise be available to the people that need to work with this system for their day to day operations. And if you make an error, by for instance forgetting to put a date filter on your query, you could even bring the system down so no one can use it anymore.
  • You can’t compare the data with data from other sources.
    • Even when the systems are similar. Like an HR system and a payroll system that use each other to work. Data is always going to be different because it is granulated on a different level, or not all data is relevant for both systems.
  • You don’t have access to old data
    • To keep the applications at peak performance, old data, that’s irrelevant to day to day operations is deleted.
  • Data is optimized to suit day to day operations
    • And not for report making. This means you’ll have to get creative with your queries to get the data you need.

So what method should you use?

That all depends on what you need at that moment. If you need detailed information about things that are happening now, you should use OLTP.
If you are looking for trends, or insights on a higher level, you should use a Data Warehouse.

 Related References

Netezza / PureData – Casting Numbers to Character Data Type

Cast Conversion Format
Cast Conversion Format

I noticed that someone has been searching for this example on my site, so, here is a quick example of how to cast number data to a character data type.  I ran this SQL example in netezza and it worked fine.

Basic Casting Conversion Format

cast(<<FieldName>> as <<IntegerType_or_Alias>>) as <<FieldName>>

 

Example Casting Number to Character Data Type SQL

 

SELECT
—-Casting Integer to Character Data Type —————

SUBMITDATE_SRKY as  SUBMITDATE_SRKY_INTEGER
, cast(SUBMITDATE_SRKY as  char(10)) as Integer_to_CHAR
, cast(SUBMITDATE_SRKY as Varchar(10)) as Integer_to_VARCHAR
, cast(SUBMITDATE_SRKY as Varchar(10)) as Integer_to_VARCHAR
, cast(SUBMITDATE_SRKY as Nchar(10)) as Integer_to_NCHAR
, cast(SUBMITDATE_SRKY as NVarchar(10)) as Integer_to_NVARCHAR

—-Casting Double Precision Number to Character Data Type —————

, HOST_CPU_SECS as DOUBLE_PRECISION_NUMBER
, cast(HOST_CPU_SECS as  char(30)) as DOUBLE_PRECISION_to_CHAR
, cast(HOST_CPU_SECS as Varchar(30)) as DOUBLE_PRECISION_to_VARCHAR
, cast(HOST_CPU_SECS as Varchar(30)) as DOUBLE_PRECISION_to_VARCHAR
, cast(HOST_CPU_SECS as Nchar(30)) as DOUBLE_PRECISION_to_NCHAR
, cast(HOST_CPU_SECS as NVarchar(30)) as DOUBLE_PRECISION_to_NVARCHAR

—-Casting Numeric to Character Data Type —————

, TOTALRUNTIME  as NUMERIC_FIELD
, cast(TOTALRUNTIME as  char(30)) as NUMERIC_FIELD_to_CHAR
, cast(TOTALRUNTIME as Varchar(30)) as NUMERIC_FIELD_to_VARCHAR
, cast(TOTALRUNTIME as Varchar(30)) as NUMERIC_FIELD_to_VARCHAR
, cast(TOTALRUNTIME as Nchar(30)) as NUMERIC_FIELD_to_NCHAR
, cast(TOTALRUNTIME as NVarchar(30)) as NUMERIC_FIELD_to_NVARCHAR
FROM NETEZZA_QUERY_FACT ;

Related References

IBM, IBM Knowledge Center, PureData System for Analytics,Version 7.2.1

IBM Netezza stored procedures, NZPLSQL statements and grammar, Variables and constants, Data types and aliases

IBM, IBM Knowledge Center, PureData System for Analytics,Version 7.2.1

IBM Netezza database user documentation,SQL statement grammar,Explicit and implicit casting, Summary of Netezza casting

IBM, IBM Knowledge Center, PureData System for Analytics,Version 7.2.1

IBM Netezza database user documentation ,Netezza SQL basics,Netezza SQL extensions

 

Databases – Database Isolation Level Cross Reference

Database Type Isolation Levels Cross Reference
Database And Tables

 

Here is a table quick reference of some common database and/or connection types, which use connection level isolation and the equivalent isolation levels. This quick reference may prove useful as a job aid reference, when working with and making decisions about isolation level usage.

Database isolation levels

Data sources

Most restrictive isolation level

More restrictive isolation level

Less restrictive isolation level

Least restrictive isolation level

Amazon SimpleDB

Serializable Repeatable read Read committed Read Uncommitted

dashDB

Repeatable read Read stability Cursor stability Uncommitted read

DB2® family of products

Repeatable read Read stability* Cursor stability Uncommitted read

Informix®

Repeatable read Repeatable read Cursor stability Dirty read

JDBC

Serializable Repeatable read Read committed Read Uncommitted

MariaDB

Serializable Repeatable read Read committed Read Uncommitted

Microsoft SQL Server

Serializable Repeatable read Read committed Read Uncommitted

MySQL

Serializable Repeatable read Read committed Read Uncommitted

ODBC

Serializable Repeatable read Read committed Read Uncommitted

Oracle

Serializable Serializable Read committed Read committed

PostgreSQL

Serializable Repeatable read Read committed Read committed

Sybase

Level 3 Level 3 Level 1 Level 0

 

Related References

 

What are the dashDB isolation levels?

dashDB
dashDB

 

Isolation levels are part of the ACID (Atomicity, Consistency, Isolation, Durability) paradigms in database control.  Isolation levels allow developers and user to trade-off consistency for a potential gain in performance. Therefore, it is essential to understand them and how the apply in structured Query Language(SQL).  The dashDB RDBMS has four isolations levels:

Repeatable Read (RR)

  • The repeatable read (RR) isolation level locks all the rows that an application references during a unit of work (UOW). If an application issues a SELECT statement twice within the same unit of work, the same result is returned each time. Under RR, lost updates, access to uncommitted data, non-repeatable reads, and phantom reads are not possible.
  • Under RR, an application can retrieve and operate on the rows as many times as necessary until the UOW completes. However, no other application can update, delete, or insert a row that would affect the result set until the UOW completes. Applications running under the RR isolation level cannot see the uncommitted changes of other applications. This isolation level ensures that all returned data remains unchanged until the time the application sees the data, even when temporary tables or row blocking is used.
  • Every referenced row is locked, not just the rows that are retrieved. For example, if you scan 20 000 rows and apply predicates to them, locks are held on all 20 000 rows, even if, say, only 200 rows qualify. Another application cannot insert or update a row that would be added to the list of rows referenced by a query if that query were to be executed again. This prevents phantom reads.
  • Because RR can acquire a considerable number of locks, this number might exceed limits specified by the locklist and maxlocks database configuration parameters. To avoid lock escalation, the optimizer might elect to acquire a single table-level lock for an index scan, if it appears that lock escalation is likely. If you do not want table-level locking, use the read stability isolation level.
  • While evaluating referential constraints, the dashDB might, occasionally, upgrade the isolation level used on scans of the foreign table to RR, regardless of the isolation level that was previously set by the user. This results in additional locks being held until commit time, which increases the likelihood of a deadlock or a lock timeout. To avoid these problems, create an index that contains only the foreign key columns, which the referential integrity scan can use instead.

Read Stability (RS)

  • The read stability (RS) isolation level locks only those rows that an application retrieves during a unit of work. RS ensures that any qualifying row read during a UOW cannot be changed by other application processes until the UOW completes, and that any change to a row made by another application process cannot be read until the change is committed by that process. Under RS, access to uncommitted data and non-repeatable reads are not possible. However, phantom reads are possible. Phantom reads might also be introduced by concurrent updates to rows where the old value did not satisfy the search condition of the original application but the new updated value does.
  • For example, a phantom row can occur in the following situation:
    • Application process P1 reads the set of rows n that satisfy some search condition.
    • Application process P2 then inserts one or more rows that satisfy the search condition and commits those new inserts.
    • P1 reads the set of rows again with the same search condition and obtains both the original rows and the rows inserted by P2.
  • In a dashDB environment, an application running at this isolation level might reject a previously committed row value, if the row is updated concurrently on a different member. To override this behavior, specify the WAIT_FOR_OUTCOME option.
  • This isolation level ensures that all returned data remains unchanged until the time the application sees the data, even when temporary tables or row blocking is used.
  • The RS isolation level provides both a high degree of concurrency and a stable view of the data. To that end, the optimizer ensures that table-level locks are not obtained until lock escalation occurs.
  • The RS isolation level is suitable for an application that:
    • Operates in a concurrent environment
    • Requires qualifying rows to remain stable for the duration of a unit of work
    • Does not issue the same query more than once during a unit of work, or does not require the same result set when a query is issued more than once during a unit of work

Cursor Stability (CS)

  • The cursor stability (CS) isolation level locks any row being accessed during a transaction while the cursor is positioned on that row. This lock remains in effect until the next row is fetched or the transaction terminates. However, if any data in the row was changed, the lock is held until the change is committed.
  • Under this isolation level, no other application can update or delete a row while an updatable cursor is positioned on that row. Under CS, access to the uncommitted data of other applications is not possible. However, non-repeatable reads and phantom reads are possible.
  • Cursor Stability (CS) is the default isolation level.
  • Cursor Stability (CS) is suitable when you want maximum concurrency and need to see only committed data.
  • In a dashDB environment, an application running at this isolation level may return or reject a previously committed row value, if the row is concurrently updated on a different member. The WAIT FOR OUTCOME option of the concurrent access resolution setting can be used to override this behavior.

Uncommitted Read (UR)

  • The uncommitted read (UR) isolation level allows an application to access the uncommitted changes of other transactions. Moreover, UR does not prevent another application from accessing a row that is being read, unless that application is attempting to alter or drop the table.
  • Under UR, access to uncommitted data, non-repeatable reads, and phantom reads are possible. This isolation level is suitable if you run queries against read-only tables, or if you issue SELECT statements only, and seeing data that has not been committed by other applications is not a problem.
  • Uncommitted Read (UR) works differently for read-only and updatable cursors.
  • Read-only cursors can access most of the uncommitted changes of other transactions.
  • Tables, views, and indexes that are being created or dropped by other transactions are not available while the transaction is processing. Any other changes by other transactions can be read before they are committed or rolled back. Updatable cursors operating under UR behave as though the isolation level were CS.
  • If an uncommitted read application uses ambiguous cursors, it might use the CS isolation level when it runs. To prevent this escalation, modify the cursors in the application program to be unambiguous and/or Change the SELECT statements to include the for read-only

 

Related References

IBM dashDB

Accessing remote data sources with fluid queries on dashDB Local, Developing for federation

 

InfoSphere / Datastage – What are The support Connectors stages for dashDB?

dashDB
dashDB

In a recent discussion, the question came up concern which Infosphere Datastage connectors and/or stages are supported by IBM for dashDB.  So, it seems appropriate to share the insight gained from the question being answered.

What Datastage Connectors and/or stages are Supported for dashDB

You have three choices as to connectors, which may best meet you your needs based on the nature of your environment and the configuration chooses which have been applied:

  1. The DB2 Connector Stage
  2. The JDBC Connector stage
  3. The ODBC Stage

Related References

Connecting to IBM dashDB

InfoSphere Information Server, InfoSphere Information Server 11.5.0, Information Server on Cloud offerings, Connecting to other systems, Connecting to IBM dashDB

DB2 connector

InfoSphere Information Server, InfoSphere Information Server 11.5.0, Connecting to data sources, Databases, IBM DB2 databases, DB2 connector

ODBC stage

InfoSphere Information Server, InfoSphere Information Server 11.5.0, Connecting to data sources, Older stages for connectivity, ODBC stage

JDBC data sources

InfoSphere Information Server, InfoSphere Information Server 11.5.0, Connecting to data sources, Multiple data sources, JDBC data sources

Netezza – [SQLCODE=HY000][Native=46] ERROR: External Table : count of bad input rows reached maxerrors limit

SQL (Structured Query Language)
SQL (Structured Query Language)

While helping a customer we encountered the [SQLCODE=HY000][Native=46] ERROR, which was a new one for me. So here are a few notes to help the next unlucky soul may run into the error.

Netezza Error Reason:

  • [SQLCODE=HY008][Native=51] Operation canceled; [SQLCODE=HY000][Native=46] ERROR: External Table : count of bad input rows reached maxerrors limit

What Does the Error Mean

  • In a nutshell, it means invalid data was submitted and could not be inserted.

What To Do

  • Basically, you need to go to the Netezza logs to see why the rows were rejected and resolve input data error, then resubmit your transactions. The logs are temporary and reused, so, you need to get to them before they are overwritten.

Where Are The Data Logs

  • In linux the logs can be found in /tmp:

For nzload Methods Logs

  • /tmp/database name.table name.nzlog
  • /tmp/database name.table name.nzbad

For External Table Load Logs

  • /tmp/external table name.log
  • /tmp/external table name.bad

Related References