Archive for November, 2011

filter and access in explain plan output

Sunday, November 20th, 2011

I have found that there is often confusion on exactly what these mean. The short answer is this: The access is the where clause used to locate the rows, which are then subject to further inclusion (filtering) by the another secondary column that wasn’t the main driver in finding rows. I think the confusion... »

Posted in Database, Oracle | No Comments »

Using awk to extract bind variable values from a trace file

Tuesday, November 15th, 2011

While troubleshooting a high CPU utilization problem recently, I found that half the sessions using a given database service were using a high amount of CPU, while the other were using a normal amount. Since I knew they were each running the exact same code, the logical conclusion was that the data differed. I... »

Posted in Database, Linux, Operating Systems, Oracle | No Comments »

Help! Select queries are…filling up my redo logs?!!

Monday, November 7th, 2011

It is theoretically possible (although unlikely in a well managed system) to cause a large amount of transaction logging to occur if you do something as simple as update a row in one session, don’t commit, and repeatedly query that same row from another session. How is this possible? Let’s assume we have a... »

Posted in Database, Oracle | No Comments »

Proving Oracle’s read consistent model

Tuesday, November 1st, 2011

I was asked a good question (by a real doubter) asking me to prove Oracle’s ability to provide consistent reads. However, in the past I have been surprised at the number of fairly senior Oracle professionals that can’t come up with a simple test case. Below is such a case. We start by creating... »

Posted in Database, Oracle | No Comments »