10G OEM Grid Control Performance Overview

I have been using Oracle's enterprise management toolsets for some time.  I was around when the migration from command line to Oracle Enterprise Manager first took place.  I must admit, it was hard for me to give up the command line interface that I trusted.  Every time I entered a command using the new tool, I used the "Show SQL" button to review the SQL that was being executed.   I began using the term "SGT" or "Sissy GUI Tool" to describe Oracle's new graphical administrative tool.    I thought that real DBAs used command line! I think I'm showing my age here.

Now, like the majority of Oracle DBAs, I am an avid user of Oracle's graphical administrative toolsets.   They make day-to-day administration so much easier that I actually can't remember the last time  I used the traditional command line interface.  I am now a "SGT" proponent. 

Oracle's New Strategy for Enterprise Manager
Previous versions of Enterprise Manager focused primarily on displaying a collection of database administration tools to the end user.  The product's primary intent was to facilitate database object administration.  Don't get me wrong.  9I OEM also allowed DBAs to be proactive by providing panels to create event driven alerts.  In addition, 9iOEM also allowed administrators to run administrative and report jobs in batch against the target databases.  If you purchased the Tuning Pack, there were dozens of screens available that displayed in-depth tuning information.   Although these monitoring and tuning tools were very useful, when you compare 9I OEM to its 10G counterpart, it is blatantly obvious that Oracle has changed the way it wants us to use Enterprise Manager.

Although the efficient administration of database objects is still one of the fundamental objectives for 10G Grid Control, its no longer the primary one.  In 10G Enterprise Manager, database object administration must "share the spotlight" with the presentation of proactive feedback information.      Each target's home page (database server, application server, hardware server, listener) displays such a wide array of information that a few of the Giant Eagle DBAs have described it as "command central."

Each home page contains a wealth of information on performance, current workload, availability, diagnostics and early warning alerts and events.  The new focus of Enterprise Manger is to provide information to the DBA to help them understand what is happening in their environments.  Don't worry, Oracle still provides plenty of emphasis on database object administration.  Each target's home page contains navigation tabs that allow administrators to easily access configuration information, management tools and administrative functions.

Enterprise Manager Database Administration Functions
Although the graphical presentation has changed from 9I OEM to 10G Grid Control, we still perform administrative activities the same way we have for years.   We add files to tablespaces, add columns to tables and grant and revoke security using buttons and drop-downs.  As I stated, the administrative procedures didn't change.  We are just using a different navigation style and clicking on different buttons and drop-downs to make object changes. 

Once you learn to navigate through the tool as you would a website, Oracle10G Grid Control becomes a breeze to use.   As usual, I didn't like it when I didn't understand how to use it.  After using it for 2 or 3 hours, I was able to easily navigate through the database administration panels.  I now prefer 10G Grid Control's navigation style to previous versions.  Who says you can't teach an old DBA new tricks? 

Enterprise Manager Performance Monitoring
The story is much different with 10G Enterprise Manager's performance and monitoring architecture.  Unlike its administrative counterpart, 10G Enterprise Manager's performance tools will change the way we monitor and tune Oracle databases.  

When I first started learning how to use 10G Grid Control's new performance screens, I was somewhat intimidated by their radically different look and feel.    Learning the new interface takes time and patience.  After testing day-to-day functionality, I spent 15 or 20 hours performing mock tuning drills using Enterprise Manager.    I once again feel that 10G Grid Control's tools are far superior to any product (third party or Oracle provided) that I have ever used.  

The changes begin with the information displayed on the target's home page.   Host CPU and user workload information is predominantly displayed.  In addition, early warning performance alert information is also displayed.   The administrator is able to use the links provided with this information to drill down to panels providing more specific and detailed tuning information.

Workload Monitoring
The interesting thing to note is that once you begin to drill down through the screens, 10G Grid Control's focus is on the SQL workload and not database statistics.   This is in stark contrast to 9I OEM, which provided dozens of screens that displayed database performance stats.  10G Enterprise Manager wants us to review the SQL workload first, drill down into it and investigate the impact it has on the database before we begin looking at the database performance indicators.    In upcoming blogs, I'll highlight this change specifically.

10G Grid Control bolsters its emphasis on SQL workload monitoring by providing users with the capability of viewing past workloads.   Administrators are able to use a slider to review SQL workloads that occurred previously.  This feature will help DBAs answer the age old question, "I had a program run long yesterday, can you help?".

In future blogs, I will be paying particular attention to 10G Grid Control's SQL monitoring feature, which automatically analyzes SQL statements for performance and resource consumption.   This enhancement not only identifies poorly performing SQL statements but also make recommendations to improve them.

Proactive Performance Monitoring
Metrics are units of measurement that are set at predefined values to assess the health of the target being monitored.  These predefined values are called metric thresholds.  Thresholds are one or more values against which current statistical values are compared.  10G Grid Control divides the threshold settings into warning and critical categories and allows administrators to configure a specific set of actions that will occur when that threshold category is reached.   For example, the DBA may want to be e-mailed if a particular host's CPU utilization exceeds the warning threshold of 70% and paged if the host's CPU utilization exceeds the critical threshold of 90%.

Each target (host, database, listener, application server) has its own set of metrics that can be customized to address each application's unique performance monitoring requirements.   In addition, 10G Enterprise Manager allows you to copy the current target's metrics to other targets to facilitate the metric creation process.   Our current strategy at Giant Eagle is to set metrics for a master target and copy those metrics to all targets of the same type.   Once we become more accustomed to using Enterprise Manager, we'll consider customizing each individual target's metrics to better address its unique monitoring needs.

10G Grid Control also takes a much more proactive role in performance monitoring and reporting.  This proactive monitoring and reporting functionality is applied to all monitored targets.  10G Grid Control immediately begins monitoring all targets at default settings when they are registered.  Administrators are then able to fine-tune the threshold values for individual targets when desired.  Monitored metrics are stored and aggregated in the repository to provide performance trend analysis capabilities. Those of us who have SLA agreements with their customers will find 10G Grid Control's status reports to be of great benefit.  10G Grid Control automatically provides an availability status for the last 24 hours, 7 days or one month.  
 
Intelligent Thresholding
10G Grid Control allows administrators to review statistics for a monitored target and choose a particular time period to act as a benchmark for future performance evaluations.   The DBA reviews past statistical information, determines when performance was good and 10G Grid Control will capture the statistics for that time period.   The DBA then sets one warning and one critical threshold for each monitored statistic.  10G Grid Control will begin comparing all future statistics to the baseline.  If any of the statistics exceed their baseline threshold, a warning or critical alert message will be generated.

Advice Driven Alerts
10G Grid Control also provides advice-driven responses to generated alerts.  The administrator is able to use the advice to speed the problem determination and resolution process.  The advice may include historical information to allow administrators to perform trend analysis on a particular statistic or threshold.

Operating System Performance Information
10G Enterprise Manager places much more emphasis on the presentation of operating system and hardware specific performance information than previous versions.  Each monitored host has its own performance panel that provides information on CPU, Memory and Disk performance statistics.  We all know the effect that the operating system and hardware combination have on database performance.  You can have the most finely tuned database on the planet and it will still not perform to expectations if the hardware server/operating system environment it runs on is configured incorrectly or is experiencing a workload spike.  Host performance information is another feature that I intend to spend some dedicated blog-time on.

Next Up
In my next blog we'll cover host performance information.  

Thanks for reading!