Read Oracle RMAN 11g Backup and Recovery Online
Authors: Robert Freeman
312
Part III: Using RMAN Effectively
The Grid Control Architecture
The Grid Control architecture (see Figure 13-3) starts with the Oracle Management Service (OMS), which is the application that is deployed on the application server. The OMS collects data from registered target servers via the central agent. The agent is installed on a target server, collects information, and pushes the data to the OMS. The OMS loads the data into the repository database. The OMS then builds web pages based on the information in the repository, which can be retrieved from any browser that can hit the OMS server’s URL.
The Central Agent
The central agent is installed at each computer that you will be monitoring with Grid Control. The agent is a “dummy” software piece, meaning it cannot make any decisions on its own. It gathers data using Perl scripts and pushes that data over HTTPS to the OMS. The OMS performs any intelligence that is required and then sends an action to the agent to perform.
The agent has a relatively small footprint, from a memory perspective. However, it can be a visible player on the CPU, depending on what you are asking it to do.
The Oracle Management Service
The OMS is a deployed web application on the middle tier of the Grid Control architecture. It is constantly receiving information from agents, in the form of .xml files, that it then loads into the tables of the repository. It is responsible for building web content for the HTTP server that Central agent
Central agent
Central agent
Monitored
Monitored
Monitored
server
server
server
Oracle Management server
Central agent
Repository
database
Grid Control
server
HTTP server
FIGURE 13-3
Grid Control architecture
Chapter 13: Using Oracle Enterprise Manager for Backup and Recovery
313
provides the console web pages, and as such may ask the agent for specific information.
However, the typical data is pushed from the agent, rather than the OMS pulling it.
The Repository Database
The OMS uses an Oracle database for its data source. The repository database is used to store information about the managed targets, as well as Grid Control operations (such as jobs or notifications).
Advanced RDBMS features are put to good use within the repository; Advanced Queuing (AQ) gets a workaround, partitioning is employed heavily, and even the internal DBMS_JOB is used.
Installing and Configuring Grid Control
In the previous version of this book, we spent a few pages going through the Grid Control installation.
Because other books have caught up to Grid Control and cover this topic better, and because we will be showing you examples from here on that have to do with utilizing Enterprise Manager Database Control, we are forgoing that conversation here. The best place for comprehensive coverage of Grid Control is the Oracle Press title,
Oracle Enterprise Manager Grid Control 10
g
Handbook.
That being said, we do want to make one final point about Grid Control. Grid Control is not something you casually install somewhere. Be very careful when you are testing it, because you can easily overwhelm resources at your disposal (especially in extremely low-cost testing environments such as the ones we’ve been known to discuss in the RMAN books).
NOTE
If you do have both the Grid Control centralized agent and Database
Control running on the same server, you must go to great lengths to
keep yourself oriented as to which $ORACLE_HOME directory you
are in at each step. This is because an
emctl
executable will exist in
the bin directory of every ORACLE_HOME (including the ORACLE_
HOME that houses the agent), but bad things happen if you start
running the wrong
emctl
command in the wrong ORACLE_HOME.
We suggest running a nice set of environment-changing scripts before
any operation.
Database Control
As we stated previously, Database Control is a subset of Grid Control functionality, limited to the management and monitoring of a single database. As such, the functionality is database-centric but not exactly limited to just the database. There are host statistics gathered for the server on which the database resides, so the host, while not a separate “target,” does have some information reported. In addition, Database Control can be configured to monitor and administer an Automatic Storage Management (ASM) instance that the target database uses for storage.
Database Control can also be configured to monitor a RAC database, and thus to monitor multiple hosts, multiple instances, and multiple ASM instances. Still, it can monitor only a single database—so the limitation of Database Control is still the same.
The Database Control Architecture
The Database Control architecture is very similar to that of Grid Control, but the scale is much smaller. The central agent and the OMS are rolled into the same OC4J application, and the repository is housed in the target database itself. Figure 13-4 shows how this architecture looks.