Systems Engineering and RDBMS

Oracle 11g

Posted by decipherinfosys on February 22, 2007

Last year at their OpenWorld conference, Oracle announced their plans for Oracle 11g – the next release version for Oracle’s RDBMS.  Overall, they have promised 482 new features.  However, not much information has been released about them.  A couple of key ones that were discussed then and in subsequent discussion boards are: .

  1. A better table and index compression scheme: This is a feature that IBM also has in Viper (DB2 version 9). The segment compression will work for all DML statements and not just the direct path loads.  So, we can now create tables as compressed and use them in our regular OLTP applications with very less overhead.
  2. Capturing and replaying the database workloads: This is one feature that is present in SQL Server (profiler replays).  One can now capture the workload in production and then replay it in development.
  3. Virtual columns and virtual indexes: Virtual columns are columns that are actually functions (kind of like computed columns in SQL Server).
  4. ADDM for RAC: This will make troubleshooting RAC configurations much easier.
  5. Interval partitioning: This scheme will automatically create time based partitions as new data gets added.  This is essentially an automated version of range partitioning.
  6. RMAN Improvements: RMAN can now by-pass the UNDO tablespace.
  7. ADR: Automatic Diagnostic Repository.  Whenever critical errors will occur in the system, an incident will be automatically created and certain health checks can be run automatically.  This information gathered then can be used by Oracle Support to better help customers.
  8. Feature Based Patch-sets: All the one-off patches will be categorized by feature sets.  So, you can easily identify which patches you need based on the features that you use.
  9. Automatically tuning the bad SQLs: In the 10g version, the automatic tuning advisor can make tuning suggestions based in the form of SQL profiles.  In 11g, one can automatically apply SQL profiles for statements if the suggested profile gives atleast 3 times better performance – please note that these performance comparisons are done during a DBA specified maintenance window and needs to be carefully evaluated – we will post more details when they become available.
  10. Improvements to the Resource Manager: The resource managerin 11g can manager I/O in additon to CPU.  You can even set up priorities with files, file types or the ASM disk groups.
  11. A couple of new load balancing utilities like Oracle Data Guard load balancing between standby databases (this was actually introduced in 10gR2 itself and has been improved upon further in 11g).
  12. A new hint to cache the data in the data buffers and not the intermediate data blocks that are accessed to obtain the results for the queries.
  13. Faster collection of stats using dbms_stats.
  14. Automatic memory tuning: This is great – but I bet will not work the way it is supposed to 🙂  Oracle had introduced automatic PGA tuning in 9i version, automatic SGA tuning in 10g which still has a lot of issues.  11g promises to tune all memory by setting one parameter.  You will just need to tell Oracle how much memory is allocated to it and it will internally dynamically manage the memory for PGA, SGA and the OS processes.  This is just like SQL Server – minimum and maximum values can be set and leave the rest on Oracle.
  15. FGDT: Fine Grained Dependency Tracking: What this means is that when one adds say a cursor to a package, there will be no invalidation of the objects that depend on it.
  16. PL/SQL compiler changes: One of the changes is to issue a warning for a “when others” without a raise.
  17. Sequences: Normal assignments on sequence values would be possible and will by-pass the DML (sequence_name.nextval).

We will be writing more detailed blogs on these feature sets in the days to come as more information becomes available.  The aim of 11g release is ease of administration and lots of new programming and HA (High Availability) and Scalability features.  Also, the world of SQL Server, Oracle and DB2 LUW (in terms of feature sets) is coming closer as each vendor starts adopting the good features of another one – all this is good for us – the end consumers.

    Sorry, the comment form is closed at this time.

    %d bloggers like this: