A lot of the AWR reports ask for, before spitting out their report, the number of days back you would like to go before entering your beginning and ending snapshot IDs. When they report on all the snapshot IDs they have left out, I think, one very important piece of information that just might cloud our judgment when selecting the proper snapshot range. This would be whether, during the snapshot period, there has been a bounce of the database–reseting the statistics to zero.
This script I have here is very similar to the ones in the AWR reports ($ORACLE_HOME/rdbms/admin/awr*) but also shows when the database has been bounced during a snapshot. We can do this by the following SQL which joins the DBA_HIST_DATABASE_INSTANCE and DBA_HIST_SNAPSHOT views—showing historical information on the snapshots in the Workload Repository. We obviously need to join these tables on the dbid, instance_number, and, the important part, startup_time. We also make sure that we only bring back snapshots that are newer than the number of days back specified by the user by comparing the time of the actual snapshot (end_interval_time). Please note that this script will output a status of ‘**db restart**’ for those times that the database was down and unavailable. This is very important as it shows us those times that Oracle was not collecting statistics (the database was down) and more importantly the statistic counters were zeroed. We can report on a bounce condition if the startup_time and begin_interval_time are the same.
prompt prompt prompt Enter the number of days to look for snapshot IDs prompt ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ select dhdi.instance_name, dhdi.db_name, dhs.snap_id, to_char(dhs.begin_interval_time,'MM/DD/YYYY:HH24:MI') begin_snap_time, to_char(dhs.end_interval_time,'MM/DD/YYYY:HH24:MI') end_snap_time, decode(dhs.startup_time,dhs.begin_interval_time,'**db restart**',null) db_bounce from dba_hist_snapshot dhs, dba_hist_database_instance dhdi where dhdi.dbid = dhs.dbid and dhdi.instance_number = dhs.instance_number and dhdi.startup_time = dhs.startup_time and dhs.end_interval_time >= to_date(sysdate - &&num_days_back) order by db_name, instance_name, snap_id;
Glad to be back. It HAS been awhile and hopefully you forgive thecheapdba for staying away.
BUT this post, I am sure you will like. It is an installation guide for Oracle 11g on Linux CentOS-5.
As the installation is quite lengthy I will just provide you with a link to the main, new and “improved” website location.
So just visit http://www.thecheapdba.com/articles/Install_Oracle11gCentOS5.htm to take a look at it!
Cheers, and it won’t be this long in between posts again.
The numbers are compelling: Oracle's share is 23.2%. A cluster of five other vendors have between 9% and 14% each. The rest is spread broadly, with each vendor commanding 2% or less. Oracle's share grew 23.3%, compared to growth of just under 12% for the sector as a whole.
I am glad to see this for a bunch of reasons. As Vice President of Embedded Technology at Oracle, I take a personal interest, of course. Oracle Berkeley DB, which Oracle acquired with Sleepycat in 2006, is aimed squarely at the embedded space. I have long maintained that embedded opportunities represent a significant source of new revenue and growth. Computers have escaped the data center, and special-purpose systems are getting deployed in living rooms, in the walls of buildings and in shirt pockets. There is an enormous amount of data travelling over networks and touching these systems.
The key to our success in the embedded space has been to assemble a family of products that address a wide range of requirements. A manufacturer building mobile telephone handsets needs to store crucial information reliably. So does a vendor building an optical network switch, and an ISV developing high-performance equity trading systems for financial markets. The three have very different requirements, though, and it's unrealistic to expect any single product to satisfy all of them.
All of our database products -- Oracle Database, Oracle TimesTen, Oracle Berkeley DB and Oracle Lite -- can be embedded in partner systems and deployed invisibly to end users. All contributed to our number one ranking by IDC.
It's not just the technology that has made us successful, though. The people who choose and deploy embedded databases are software developers. In the enterprise, we generally talk to DBAs and CIOs, but in the embedded world, we talk to architects and CTOs. Those conversations are different, and we have had to develop new expertise and new strategies as we have pursued embedded customers. Over the past several years, we've concentrated on building the technical, support and sales expertise necessary to win embedded business in countries around the globe. IDC's vendor share numbers suggest that we're doing okay.
Congratulations to Oracle's Embedded Global business team, and to the product development and support groups for all four products! This is a tremendous accomplishment.