Archive for July, 2013|Monthly archive page

No Data and Big SQL: Oracle Database 12c on Exadata and SuperCluster

In DBA, Exadata, Oracle database on July 25, 2013 at 10:51

“If you want to succeed you should strike out on new paths, rather than travel the worn paths of accepted success.” John D. Rockefeller

No Data: with Exadata, with the help of storage indexes, you can sometimes perform full table scans without any physical I/O! Really, no data is being read or written to! Richard Footie explains the details in his article entitled Exadata Storage Indexes – Part I (Beginning To See The Light).

Big SQL: with Oracle Database 12c, on a single AWR container report, you can see SQLs from all databases (up to 252) plugged into the container databases, that is a big SQL report! We will see lot of Big SQL with the increased usage of container databases. You can see a screenshot of a 12.1.0 AWR report (SQL ordered by Elapsed Time) at the end of this blog post. Note that in IBM terminology, Big SQL is the SQL interface to its Hadoop-based platform, InfoSphere BigInsights.

How about 12c on Exadata or SuperCluster? Is it supported?


Yes, Oracle 12.1.0 is supported on Exadata and SPARC SuperCluster. The system should run Exadata software version You can have both 11gR2 and 12cR1 databases running simultaneously on the same cluster.

The Grid Infrastructure must be first upgraded from 11g to 12c. You may install fresh 12c GI only in case you do not plan on creating new 11gR2 databases (this will fail with SSO wallet incompatibility due to Bug 16844086).

PDB Quiz

Oracle Database 12c running on Exadata or SuperCluster requires the following:

• Exadata version or later installed on Exadata Storage Servers and database servers
• Patch 16547261 applied on Exadata Storage Servers
• Database servers running Oracle Linux Unbreakable Enterprise Kernel (UEK) 2.6.32-400.21.1.el5uek updated to 2.6.32-400.29.1.el5uek. Other Oracle Linux kernels supplied with Exadata do not require update.

Note that Oracle Exadata is sometimes still being referred to as Oracle SAGE (SAGE = Storage Appliance for Grid Environments):


Oracle Database 12c running on Exadata Database Machine is subject to the following restrictions:

• The 12.1 offload libraries are not present on Exadata version Therefore, Smart Scan offloaded filtering and storage indexes are disabled for 12.1 databases. Smart Scans
are still issued, and the initiation of IOs is offloaded, but complete blocks are returned instead of just the selected rows and columns. 11.2 databases on the same system have
full offload capabilities. A future release of Exadata storage server software will add 12.1 offload libraries and lift these restrictions.

• I/O Resource Management (IORM) plans for 12.1 database are not enforced.
– Database resource plans (intradatabase resource plans) set on 12.1 databases are not enabled on cells.
– An interdatabase plan cannot manage 12.1 databases, therefore an interdatabase plan cannot be used to manage flash log or flash cache for a 12.1 database.

• Cell metrics for 12.1 databases are reported under OTHER_DATABASE.

Other Exadata features like smart flash cache, smart flash log, hybrid columnar compression, and low latency and high throughput infiniband messaging continue to work with 12.1 databases. Note that these restrictions only effect performance of 12.1 databases. All 12c database functionality, such as Pluggable Databases, is fully available.

Not directly related to the topic but look at the AWR report in 12c. It shows the PDB name where the SQL was run:


AWR reports in 12.1.0 are slightly different than those in 11g. CDBs and PDBs are the main factor.

Back to Exadata: the new Oracle Exadata Plug-in comes with several new features. Now, it supports SPARC SuperCluster, 1/8 Rack and multi-rack certifications. Check the complete list here.