David Fitzjarrell

I don’t know what auditing requirements may be in place in other parts of the world, but here in the U.S. we have the Sarbanes-Oxley Act (resulting from the Enron debacle from a few years back) which basically calls for audit trails for all financial activity in an organization, to prevent such cover-ups from recurring. […]

Smart Scans are part of what Kerry Osborne calls the ‘secret sauce’ of Exadata, and he’s right. Smart Scans can definitely improve throughput by using the various optimizations to reduce the amount of data read and the volume of data returned to the database servers. You really want to have them working on an Exadata […]

An oft-used (and subsequently oft-abused) execution path is parallel execution, usually ‘instigated’ by some sort of parallel hint. Developers, albeit with the best intentions, misuse and abuse this mechanism because of faulty logic, that ‘logic’ being that if one process executes in X amount of time then Y parallel processes will execute in X/Y amount […]

A fairly common question recently appeared in an Oracle-related newsgroup; the question was titled ‘index didn’t work’ and was presented thus: “a table TEST,includes two columns:c_1(number not null),c_2(varchar2(255));10 rows inserted;create independent index on each column,then i query:select c_1 from TEST,explain plan report:index full scan,that’s right.next,i query:select c_2 from TEST,explain plan report:table access full.why?” Okay, that’s […]

Performance is the reason companies choose Exadata over other platforms, as it can provide avenues of improvement commodity hardware can’t match.  One of those avenues is the Smart Scan, a mechanism where the database servers and the storage cells join forces to return the requested results.  Let’s see what triggers a Smart Scan and discover […]

Exadata Storage Indexes Exadata offers many features to improve performance — Smart Scans, offloading, Infiniband internal network — but the one feature not often mentioned is the storage index. Designed to prevent Exadata from reading data blocks unnecessarily, its purpose is sometimes not clearly explained, leading to confusion. So what is a storage index, where […]

Due to an unfortunate series of events I was unable to attend Oracle OpenWorld this year, the year when Oracle 12c was announced to be available after the first part of 2013. I did, however, have contacts at the conference and can report on several of the new features offered in Oracle Release 12c. Let’s […]

In 2008 Oracle introduced Exadata, and data warehousing, business intelligence and OLTP applications have benefited from this integrated technology stack.  Exadata is not like anything most DBAs have ever seen or managed; it’s a tuned combination of database server, storage server and private interconnects that truly can outperform commodity hardware configurations.  Let’s take a peek […]

An interesting question was posed on the Oracle-L list a few weeks back; a member was having difficulty getting a constraint created using an existing non-unique index, as Oracle would consistently consume all of the TEMP space and cause the process to terminate for lack of resources. The conventional wisdom was that since that index […]

In version 9.2 Oracle changed the way the SGA allocates memory by implementing subpools; each subpool contains all of the necessary memory structures so this allows sessions to access shared areas without incurring the latching waits earlier releases experienced. Unfortunately version 9.2.0.1 through 9.2.0.4 could still throw ORA-04031 errors when a subpool filled, as Oracle […]

Oracle 11g offers a new twist on triggers, the compound trigger, a trigger that can act both before and after an update, insert or delete has occurred. This makes possible the ability in one trigger to perform processing similar to a stored procedure without having to write such a procedure to call from a traditional […]

AWR baselines have been available since Oracle 10g; Oracle 11g enhances those baselines with new ways to generate them and new tricks in OEM to use adaptive thresholds to monitor the performance of the database. Starting off let’s look at the way generating baselines has changed. The original 10g implementation still exists, with a new […]