Transactions in J2EE

Some resources for a good introduction to Transactions in J2EE :

• A Guideline to Transaction Management in J2EE Applications by Debu Panda – July 08, 2004
• XA Exposed III.I by Mike Spille – March 09, 2004
• XA Exposed III by Mike Spille – January 29, 2004
• XA Exposed, Part II by Mike Spille – January 13, 2004
• XA Exposed, Part I by Mike Spille – January 05, 2004
• XA Exposed: How 2 Phase Commit works in J2EE by Dion Almaer – TheServerSide – January 05, 2004
• Transactions – J2EE 1.4 Tutorial – November 16, 2003
• Transactional services in WebSphere Application Server Enterprise V5 by Peter Kovari and Daniel Cerecedo Diaz – IBM Redbook – September 24, 2003
• Transaction in J2EE by Peter Kovari and Jan Smolenski – IBM Redbook – June 11, 2003
• [FR] Les transactions dans le contexte J2EE – Objectiva – December 05, 2002
• Transactions – J2EE 1.3 Tutorial – April 24, 2002
• Understanding JTA – the Java Transaction API (cont.) – DataDirect – 2002 NEW
• J2EE Transaction Frameworks: Distributed Transaction Primer by Dibyendu Baksi – 05/23/2001 NEW
• J2EE Transaction Frameworks: Building the Framework by Dibyendu Baksi – 04/26/2001 NEW
• Transaction management under J2EE 1.2 by Sanjay Mahapatra – Java World – July, 2000
• Using Database Transactions in Java by Brett Spell – Java Pro Magazine – September, 1999

Advertisement

Transation Isolation Levels

Discussing with a colegue about defining transactions in our application we finished by talking about transaction isolation levels and finally I got into problems explaining the problems faced in concurrent transactions (I wasn’t able to tell the difference between non-repeatable and phantom reads [sigh/]). I decided to write down for further references.

Transaction isolation levels are defined according to the state of data during the following 3 scenarios (they compair the data accessed from 2 different transactions):
* dirty reads
occur when a transaction reads data written by concurrent uncommited transaction;
* non-repeatable reads
occur when a transaction re-reads data it has previously read and finds that data has been
modified by another transaction (that committed since the initial read);
* phantom read
occur when a transaction re-executes a query returning a set of rows that satisfy a search
condition and finds that the set of rows satisfying the condition has changed due to another
recently-committed transaction;

Note: even if it seems that non-repeatable reads and phantom reads are the same thing, the difference is made from the fact that in the 1st case the same rows are returned but their data is modified, while in the 2nd case another set of rows is returned.
Defined isolation levels:

DB Isolation Level	Dirty Read	Non-Repeatable Reads	Phantom read
Read uncommitted	Y	Y	Y
Read committed	N	Y	Y
Repeatable read	N	N	Y
Serializable	N	N	N

Java Level	Dirty Read	Non-Repeatable Reads	Phantom read
TRANSACTION_NONE	Y	Y	Y
TRANSACTION_READ_UNCOMMITTED	Y	Y	Y
TRANSACTION_READ_COMMITED	N	Y	Y
TRANSACTION_REPEATABLE_READ	N	N	Y
TRANSACTION_SERIALIZABLE	N	N	N

Still I would like to see more details on the separation made on non-repeatable reads and phantom reads.

Tomcat – JRockit

Found this on this weblog: Random thoughts:

Turns out that it wasn’t Tomcat 5 that provided the 15-20ms/page performance. Instead, it was the JRockit 8 JVM that did the trick. I got the same performance with Tomcat 4, but switching to Sun JVM /w server settings increased the rendering time to about 40ms/page. Oh well.

I definitely must try this.