Enterprise Manager 10g Wrangles Oracle Wares
Oracle Enterprise Manager 10g Grid Control R4 is a high-powered ecosystem
management platform that uses its home field advantage in Oracle shops to
provide administrators with top-notch tools for performance monitoring,
alerting and reporting, and job and compliance management of the Oracle systems
and applications they oversee.
Based on my tests, I recommend that administrators consider a management strategy that brings Enterprise Manager in over time to take care of Oracle databases, application servers, Web applications from Oracle and its fleet of acquired products from PeopleSoft, Siebel, JD Edwards and others. For organizations that depend on Oracle products there probably isn't a better collection of tools for monitoring the overall health and well-being of the collective services and systems needed to keep business running. However, as with most Oracle solutions, this functionality doesn't come cheap.
While the basic Enterprise Manager 10g Grid Control R4 product is available at no extra license cost with most Oracle products, the various add-ons available for the product can quickly add to the solution's costs. For instance, the configuration I tested included Oracle's Application Server Diagnostics Pack, which is priced at $6,000 per CPU, and Oracle's Management Packs for Business Intelligence and Data Masking, which are priced at $10,000 per CPU apiece. I also tested Oracle's Root Cause Analysis Management Connector, which provides information about the possible causes of service failures, and is priced at $5,000 per CPU.
Oracle also offers a Diagnostics Pack for Non-Oracle Middleware, at a cost of $6,000 dollars per CPU. I did not test the product with any non-Oracle services, but I plan to do so for a future story. For shops with an intensive Oracle focus, folding management of non-Oracle resources such as IBM's DB2 and Microsoft's SQL Server databases, Citrix's Presentation Server or any of the various Microsoft servers that Enterprise Manager supports offers the attractive prospect of keeping tabs on one's infrastructure while cutting back on the "panes of glass" required to do the job. For more information on the services that Oracle's Enterprise Manager supports, see Oracle's data sheet, here.
Enterprise Manager in the lab
During my tests of Enterprise Manager, I set up user accounts and notifications, ready-to-use and customized monitors, snapshot and baseline performance metrics, and performance alerts and reports, all aimed at getting a handle on the Oracle Database 10g and Oracle Application Server environment that I assembled for this test. I used the 10g version of Oracle's database, but the Enterprise Manager can also manage Oracle's new 11g database version, despite what the product's (somewhat confusing) naming convention might suggest.
After configuring my system monitors for my Oracle Database 10g and Oracle Application Servers, I spent most of my time creating service-level models to define the operation of individual services, or of logical groups that together provide discrete services. For instance, I configured monitoring for a listener and a group of hosts that together provided a Web-hosted application.
Even administrators who aren't already familiar with Enterprise Manager (the product was first released in 2005 and is available as a free trial on DVD by contacting Oracle) should have little trouble navigating the product's tools for discovering and classifying supported management targets. What's more, because Enterprise Manager comes with myriad preconfigured monitors, initial setup of the product will likely depend more on the time it takes to go through the normal test and proof cycle used by prudent administrators than any real learning curve in coming up to speed on Enterprise Manager.
One of my main requirements of any management tool is that it efficiently identify the root cause of any problem in the systems and applications under my command. False positives, incorrect identifications of a normal situation as a problem, are one enemy of efficiency. Flooding negatives, large numbers of correctly identified but unnecessarily repeated alerts, are another. In my testing of Enterprise Manager I encountered very few examples of either of these conditions. Granted, my test environment was rather small, and the activity level of my test applications is different from actual production environments. With that said, I credit much of the efficiency in problem identification that I experienced in my tests to the product's well-thought-out service-level modeling and monitoring logic.
For example, I was able to create service-level monitors that accommodated varying levels of system usage throughout the day, which allowed predictable production spikes to occur without setting off an alert to the management console. Conversely, when I forced database usage spikes beyond acceptable limits, these same performance monitors sent out the appropriate alerts.