This chapter explains what multithreaded agents are, how they contribute to the overall efficiency of a distributed database system, and how to administer multithreaded agents.
This chapter contains the following sections:
Administering Multithreaded Agents
Note: Heterogeneous Services supports multithreaded agents, but this functionality may not be available in all Heterogeneous Services gateways. In addition, multithreaded agents support only Oracle Transparent Gateways, not Generic Connectivity Agents like ODBC. Refer to your Transparent Gateway documentation to determine if your gateway is certified to work as a multithreaded agent. |
See Also: Oracle Database Application Developer's Guide - Fundamentals for information about configuring theextproc agent to run in multithreaded agent (MTA) mode |
This section explains how multithreaded agents contribute to the efficiency of Heterogeneous Services and Oracle Transparent Gateways.
This section contains the following topics:
By default, a Heterogeneous Services agent is started for each user session. When a user session attempts to access a non-Oracle system through a database link, an agent process is started that is dedicated to that user session and that database link. The agent process terminates only when the user session ends or when the database link is closed. Separate agent processes are started under the following conditions:
The same user session uses two different database links to connect to the same non-Oracle system
Two different user sessions use the same database link to access the same non-Oracle system
This architecture may consume an unnecessarily large amount of system resources. For example, suppose that there are several thousand user sessions simultaneously accessing the same non-Oracle system. Because an agent process is started for each session, there are several thousand agent processes running concurrently. The agent processes operate regardless of whether each individual agent process is actually active at the moment. Thus agent processes and open connections can consume a disproportionate amount of system resources. When sessions connect to the Oracle database server, this problem is addressed by starting the server in shared server mode. Shared server mode allows database connections to be shared by a small number of server processes.
The Oracle shared server architecture assumes that even when there are several thousand user sessions currently open, only a small percentage of these connections are active at any given time. In shared server mode, there is a pool of shared server processes. User sessions connect to dispatcher processes that place the tasks requested in a queue. The tasks are picked up by the first available shared server processes. The number of shared server processes is usually less that the number of user sessions.
Multithreaded Heterogeneous Services agents provide similar functionality for connections to non-Oracle systems. The multithreaded agent architecture uses a pool of shared agent threads. The tasks requested by the user sessions are put in a queue and are picked up by the first available multithreaded agent thread. Because only a small percentage of user connections are active at a given moment, using a multithreaded architecture allows for more efficient use of system resources.
One multithreaded agent must be started for each system identifier (SID) before any attempt is made to connect to the non-Oracle system. This is done using the agent control utility agtctl
. This utility is also used to configure the agent and to shut down the agent.
Each Oracle Net listener that is running on a system listens for incoming connection requests for a set of SIDs. If the SID in an incoming Oracle Net connect string is one of the SIDs that the listener is listening for, then that listener processes the connection. Further, if a multithreaded agent has been started for the SID, then the listener passes the request to that agent.
In the architecture for multithreaded agents, each incoming connection request is processed by different kinds of threads:
A single monitor thread. The monitor thread is responsible for the following:
Maintaining communication with the listener
Monitoring the load on the process
Starting and stopping threads when required
Several dispatcher threads. The dispatcher threads are responsible for the following:
Handling communication with the Oracle server
Passing task requests to the task threads
Several task threads. The task threads handle requests from the Oracle processes.
The multithreaded agent architecture is illustrated in Figure 5-1. Each request issued by a user session is represented by a separate type of arrow. There is no representation of the monitor thread in this illustration, because the monitor thread is created once when the multithreaded agent is started, and it creates and monitors the other threads. There are usually many more task threads than dispatcher threads.
These three thread types roughly correspond to the Oracle multithreaded server PMON, dispatcher, and shared server processes, respectively.
Note: All requests from a user session go through the same dispatcher thread, but can be serviced by different task threads. It is also possible for several task threads to use the same connection to the non-Oracle system. |
Each type of thread is discussed in more detail in the following sections:
See Also: Section 5.3, "Administering Multithreaded Agents" for more information about starting and stopping multithreaded agents by using the agent control utility |
When the agent control utility starts a multithreaded agent for a SID, the agent control utility creates the monitor thread. The monitor thread performs the following functions:
It creates the dispatcher and task threads.
It registers the dispatcher threads with all the listeners that are handling connections to this agent. While the dispatcher for this SID is running, the listener does not start a new process when it gets an incoming connection. Instead, the listener hands over the connection to this same dispatcher.
It monitors the other threads and sends load information about the dispatcher threads to all the listener processes handling connections to this agent. This enables the listeners to hand over incoming connections to the least loaded dispatcher.
It continues to monitors each of the threads it has created.
Dispatcher threads perform the following functions:
They accept incoming connections and task requests from Oracle servers.
They place incoming requests on a queue for a task thread to pick up.
They send results of a request back to the server that issued the request.
Note: After a user session establishes a connection with a dispatcher, all requests from that user session go to the same dispatcher until the end of the user session. |
One multithreaded Heterogeneous Services agent must be started for each system identifier (SID) before any attempt is made to connect to the non-Oracle system. Any agent not started in this fashion will not function in multithreaded mode and must be set up as described in Section 4.1, "Setting Up Access to Non-Oracle Systems".
A multithreaded agent is started, stopped, and configured by an agent control utility called agtctl
, which works like lsnrctl
. However, unlike lsnrctl
, which reads a configuration file (listener.ora
), agtctl
takes configuration information from the command line and writes it to a control file.
The following topics are discussed in this section:
You can start and stop agtctl
and create and maintain its control file by using the commands shown in Table 5-1.
Table 5-1 Agent Control Utility Commands
Command | Description |
---|---|
startup |
Starts a multithreaded agent |
shutdown |
Stops a multithreaded agent |
set |
Sets a configuration parameter for a multithreaded agent |
unset |
Causes a parameter to revert to its default value |
show |
Displays the value of a configuration parameter |
delete |
Deletes the entry for a particular SID from the control file |
exit |
Exits shell mode |
help |
Lists available commands |
These commands can be issued in one of two ways:
You can issue commands from the UNIX or DOS shell. This mode is called single-line command mode.
You can enter agtctl
and an AGTCTL>
prompt appears. You then can enter commands from within the agtctl
shell. This mode is called shell mode.
The syntax and parameters for agtctl
commands depend on the mode in which they are issued.
Note:
|
This section describes the use of agtctl
commands. They are presented in single-line command mode.
Set the configuration parameters for a multithreaded agent before you start the agent. If a configuration parameter is not specifically set, a default value is used. Configuration parameters and their default values are shown in Table 5-2.
Use the set
command to set multithreaded agent configuration parameters.
Syntax
agtctl set parameter parameter_value agent_sid
parameter
is the parameter that you are setting.parameter_value
is the value being assigned to that parameter.agent_sid
is the SID that this agent will service. This must be specified for single-line command mode.
Example
agtctl set max_dispatchers 5 salesDB
Use the startup
command to start an agent in multithreaded mode.
Syntax
agtctl startup agent_name agent_sid
agent_name
is the name of the agent executable. For example, extproc
is an agent name.agent_sid
is the SID that this agent will service. This must be specified for single-line command mode.
Example
agtctl startup extproc salesDB
Use the shutdown
command to stop a multithreaded agent. There are three forms of shutdown:
Normal
Normal shutdown is the default. It causes agtctl
to talk to the agent and ask it to terminate itself gracefully. All sessions complete the operations they are currently doing and then shut down.
Immediate
For an Immediate shutdown, agtctl
talks to the agent and tells it to terminate immediately. The agent process exits immediately regardless of the state of current sessions.
Abort
For an Abort shutdown, agtctl
does not talk to the agent at all. It issues a system call to kill the agent process.
Syntax
agtctl shutdown [immediate|abort] agent_sid
agent_sid
is the SID that the agent services. It must be specified for single-line command mode.
Example
agtctl shutdown immediate salesDB
To examine the value of a configuration parameter, use the show
command.
Syntax
agtctl show parameter agent_sid
parameter
is the parameter that you are examining.agent_sid
is the SID that this agent will service. This must be specified for single-line command mode.
Example
agtctl show max_dispatchers salesDB
You can reset a configuration parameter to its default value using the unset
command.
Syntax
agtctl unset parameter agent_sid
parameter
is the parameter that you are modifying (or unsetting).agent_sid
is the SID that this agent services. It must be specified for single-line command mode.
Example
agtctl unset max_dispatchers salesDB
In shell mode, start agtctl
by entering the following:
agtctl
This results in an AGTCTL>
prompt. Thereafter, because you are issuing commands from within the agtctl
shell, you do not need to prefix the command string with agtctl
.
Set the name of the agent SID by entering the following:
AGTCTL> set agent_sid agent_sid
All subsequent commands are assumed to be for the specified SID until the agent_sid
value is changed. Unlike single-line command mode, you do not specify agent_sid
in the command string.
You can set the language for error messages as follows:
AGTCTL> set language language
The commands themselves are the same as those for the single-line command mode. To exit shell mode, enter exit
.
The following are examples of shell mode commands.
Table 5-2 lists the configuration parameters for the agent control utility.
Table 5-2 Initialization Parameters for agtctl
Parameter | Description | Default Value |
---|---|---|
max_dispatchers |
Maximum number of dispatchers | 1 |
tcp_dispatchers |
Number of dispatchers listening on tcp (the rest are using ipc) | 0 |
max_task_threads |
Number of task threads | 2 |
max_sessions |
Maximum number of sessions | 5 |
listener_address |
Address on which the listener is listening (needed for registration) |
(ADDRESS_LIST=
(ADDRESS=
(PROTOCOL=IPC)
(KEY=PNPKEY))
(ADDRESS=
(PROTOCOL=IPC)
(KEY=listener_sid))
(ADDRESS=
(PROTOCOL=TCP)
(HOST=127.0.0.1)
(PORT=1521)))
Note: listener_sid is the IPC key to where the listener is listening of the Oracle database. |
shutdown_address |
Address the agent uses to communicate with the listener. This is the address on which the agent listens for all communication, including shutdown messages from agtctl |
(ADDRESS= (PROTOCOL=IPC) (KEY=listener_sid || agent_sid)) (ADDRESS= (PROTOCOL=TCP) (HOST=127.0.0.1) (PORT=1521))Notes:
|