You can achieve client high availability and scalability in Oracle Real Application Clusters (Oracle RAC) using database services, Fast Application Notification (FAN), Fast Connection Failover (FCF), and Application Continuity. Older applications can use Transparent Application Failover (TAF).
You can deploy Oracle RAC environments to achieve continuous service in many different ways.
Applications using a clustered database generally want to load balance their workload across the cluster. Oracle Real Application Clusters (Oracle RAC) runs on Oracle Clusterware, which provides a highly available (HA) application framework. Oracle Clusterware provides the necessary service and integration points between Oracle RAC and custom enterprise applications using FAN. Across data centers Global Data Services (GDS) also provides these integration points for services and FAN.
Depending on the number of nodes and your environment's complexity and objectives, your choices for the optimal workload management and high availability configuration depend on a variety of considerations.
To implement continuous service for your applications using Oracle RAC database, use the following features:
UP
or DOWN
events.See Also:
Services divide workloads into mutually disjoint groupings. Each service represents a workload with common attributes, service-level thresholds, and priorities.
Dynamic database services, often referred to as simply services, are logical abstractions for managing workloads in Oracle Database.A single service can represent an application, multiple applications or a subset of a single application. For example, the Oracle E-Business suite defines a service for each responsibility, such as general ledger, accounts receivable, order entry, and so on. A single service can be associated with one or more instances of an Oracle RAC database, and a single instance can support multiple services.
Note:
A database service can only be offered on a single network.
Services provide the following benefits:
Provide a single entity for managing applications that compete for the same resources
Enable each workload to be managed as a unit
Hide the complexity of the cluster from the client
To manage workloads, you can define services that you assign to a particular application or to a subset of an application's operations. You can also use services to manage the workload for different types of work. For example, online users can use one service while batch processing can use a different service and reporting can use yet another service type.
Traditionally an Oracle database provided a single service and all users connected to the same service. A database always has this default database service that is the database name. This service cannot be modified. It should only be used for administrative tasks because it always enables you to connect to the database. The default database service should not be used for high availability because it cannot be disabled or relocated. Always use user-defined database services for your applications.
Note:
Do not use the default database service for application workloads because the default database service is for administration purposes and cannot be modified. The default database service has the same name as the DB_NAME
or DB_UNIQUE_NAME
database initialization parameter. Create at least one service as described in "Creating Services".
When a user or application connects to a database, Oracle recommends that you use a service for the connection. Oracle Database automatically creates one database service when the database is created. For basic or administrative connections, this may be all you need. However, for more flexibility in the management of the applications connecting to the database and their workload, you should create one or more application services and specify which database instances offer the services.
You can define services for both policy-managed and administrator-managed databases.
Policy-managed database: When you define services for a policy-managed database, you assign the service to a server pool where the database is running. You can define the service as either uniform (running on all instances in the server pool) or singleton (running on only one instance in the server pool).
Administrator-managed database: When you define a service for an administrator-managed database, you define which instances normally support that service.These are known as the PREFERRED
instances. You can also define other instances to support a service if the preferred instance fails. These are known as AVAILABLE
instances. A service that runs on an administrator-managed database must have at least one PREFERRED
instance.
Services are integrated with the Database Resource Manager, which enables you to restrict the resources that a service uses within an instance. In addition, Oracle Scheduler jobs can run using a service, as opposed to using a specific instance.
See Also:
Oracle Database 2 Day DBA for more information about client connections to the database
When configuring service failover in administrator-managed databases, you must configure preferred and available instances.
During normal operation a service can be running on any combination of the preferred and available instances, up to the cardinality of the service (the number of PREFERRED
instances defined). Only at initial startup of the service will Oracle Clusterware attempt to start the service on the PREFERRED
instances. If an instance fails, then the service fails over to one of the combined list of preferred and available instances that is not offering the service. You can also manually relocate the service to one of the instances in the combined list of preferred and available instances not offering the service.
If a service fails over to an available instance, then the service is not moved back to its preferred instance automatically. However, you can automate the relocation of a service to its preferred instance by using a FAN callout.
If you configure a preferred instance for a service, but do not specify at least one available instance for the service, then the service does not failover to another instance if the preferred instance fails.
Using Enterprise Manager, you can also specify an instance as Not Used. This setting means the service does not run on the instance, even if the preferred instance for the service fails.
See Also:
FAST APPLICATION NOTIFICATION WITH ORACLE DATABASE 12C white paper for an example of a callout script in the section "Appendix D Sample Callout program"
Oracle Database High Availability Overview for more information about the high availability products and features available for Oracle Database
Service failover in policy-managed databases behaves differently if the service is UNIFORM or SINGLETON.
When you specify that a service is UNIFORM, Oracle Clusterware attempts to ensure that the service always runs on all the available instances for the specified server pool. If the instance fails, then the service is no longer available on that instance. If the cardinality of the server pool increases and an instance is added to the database, then the service is started on the new instance. You cannot manually relocate the service to a specific instance.
When you specify that a service is SINGLETON, Oracle Clusterware attempts to ensure that the service always runs on only one of the available instances for the specified server pool. If the instance fails, then the service fails over to a different instance in the server pool. You cannot specify which instance in the server pool the service should run on.
For SINGLETON services, if a service fails over to a new instance, then the service is not moved back to its original instance when that instance becomes available again.
See Also:
Oracle Real Application Clusters Administration and Deployment Guide for more information about automatic workload management using database services
When you define a service, you can also define the management policy for that service.
You can choose either an automatic or manual management policy:
Automatic: The service always starts when the database starts.
Manual: Requires you to start the service manually after the database starts.
Note:
When you use automatic services in an administrator-managed database, during planned database startup, services may start on the first instances that become available rather than their preferred instances.
Related Topics
The Database Resource Manager is a database feature you can use to control the database resources allocated to users, applications, and services.
The Database Resource Manager ensures that users, applications, and services receive their share of the available database resources. The Database Resource Manager enables an Oracle RAC database running on one or more nodes to support multiple applications and mixed workloads with optimal efficiency.
The Database Resource Manager provides the ability to prioritize work within an Oracle database or your Oracle RAC environment. For example, high priority users, such as online workers, would get more resources to minimize response time, while lower priority users, such as batch jobs or reports, would get fewer resources, and could take longer to run. Database Resource Manager enables more granular control over resources.
Resources are allocated to users according to a resource plan specified by the database administrator. The following terms are used in specifying a resource plan:
A resource plan specifies how the resources are to be distributed among various users based on resource consumer groups.
Resource consumer groups enable the administrator to group user sessions by resource requirements. Resource consumer groups are different from user roles; one database user can have different sessions assigned to different resource consumer groups.
Resource allocation methods are the methods or policies used by the Database Resource Manager when allocating for a particular resource. Resource consumer groups and resource plans use the resource allocation methods. The database provides the resource allocation methods that are available, but the DBA determines which method to use.
Resource plan directives are a means of assigning consumer groups to particular plans and partitioning resources among consumer groups by specifying parameters for each resource allocation method.
Subplans, which the DBA can create within a resource plan, enable further subdivision of resources among different users of an application.
Levels provide a mechanism to specify distribution of unused resources among available users. Up to eight levels of resource allocation can be specified.
The Database Resource Manager enables you to map a resource consumer group to a service so that users who connect using that service are members of the specified resource consumer group, and thus restricted to the resources available to that resource consumer group.
You can access a tutorial for Resource Manager in Enterprise Manager. Simply navigate to the Cluster Database Home page, then, from the Administration menu, select Resource Manager, and then Getting Started. See Oracle Database 2 Day DBA for details on logging in to Oracle Enterprise Manager.
See Also:
Oracle Database Administrator’s Guide for more information about the Database Resource Manager
The Oracle RAC high availability framework enables Oracle RAC to maintain the database, components, and applications in a running state always.
If an instance, component, or application fails, then it can be automatically restarted to keep Oracle Database operating at full capacity.
Oracle Database focuses on maintaining service availability. In Oracle RAC, Oracle services are designed to be continuously available with workloads shared across one or more instances. The Oracle RAC high availability framework maintains service availability by storing the configuration information for each service in the Oracle Cluster Registry (OCR). Oracle Clusterware recovers and balances services across instances based on the service definition.
See Also:
Oracle Database High Availability Overview for information about determining the high availability requirements for your system
Fast Application Notification (FAN) is a high availability notification mechanism that Oracle RAC uses to notify other processes about cluster configuration and service-level information, including status changes such as UP
or DOWN
events.
One of the main requirements of a highly available application is for it to be quickly notified when something happens to critical system components. Notification enables the application to execute event-handling programs. The timely execution of such programs minimizes the time it takes to react to cluster resource organizations and the impact of cluster component failures by avoiding costly connection time outs and application time outs.
FAN enables the automated recovery of applications when cluster components fail. For cluster configuration changes, the Oracle RAC high availability framework publishes a FAN event immediately after a change occurs regarding the state of the instances in the cluster. Instead of waiting for the application to query the database and detect a problem, applications can receive FAN events and react immediately.
FAN UP
and DOWN
events can apply to instances, services, and nodes. FAN also publishes Load Balancing Advisory events. FAN UP
and DOWN
events provide the following benefits:
For DOWN
events, the disruption to the application can be minimized because sessions that are connected to the failed instance or node can be terminated. Incomplete transactions can be terminated and the application user notified immediately. Application users who request connections are directed to instances that are started and are providing the requested service.
For UP
events, when services and instances are started, new connections can be created so that the application can immediately take advantage of the extra resources.
Oracle Clusterware and Oracle RAC use Oracle Notification Service (ONS) to propagate FAN messages both within the Oracle cluster and to client or mid-tier machines. ONS is installed with Oracle Clusterware and the resources to manage the ONS daemons are created automatically during the installation process. ONS daemons run on each node of the cluster and send and receive messages from a configured list of nodes where other ONS daemons are active; this list of nodes can include nodes outside the cluster, such as application server tiers or client nodes.
FAN can also be used with Oracle Data Guard, Active Data Guard, Oracle WebLogic Server Active GridLink for RAC, Universal Connection Pool (UCP) clients, Global Data Services, and OCI-based clients including OCI/OCCI, ODP.NET and OCI Session Pool, among others.
FAN callouts are server-side executable files that Oracle RAC runs immediately when high availability events occur.
A callout is essentially a shell script or precompiled executable written in any programming language. Some examples of how you can use FAN callouts to automate the actions performed when events occur in a cluster configuration are as follows:
Starting and stopping server-side applications
Relocating low-priority services when high-priority services come online
Sending text or numeric messages to pagers
The executable files for FAN callouts are stored in the Grid_home/racg/usrco
subdirectory. If this subdirectory does not exist in your Grid home, then you must create this directory with the same permissions and ownership as the Grid_home/racg/tmp
subdirectory.
All executables in the Grid_home/racg/usrco
subdirectory are executed immediately, in an asynchronous fashion, when a FAN event is received through the Oracle Notification Service (ONS). For most event types, the callout is executed on one node in the cluster (the node generating the event), thus a copy of the executable files used by FAN callouts should be available on every node that runs Oracle Clusterware. An example of a callout script can be found in the section "Appendix D Sample Callout program (PERL based)" in the following white paper, available on Oracle Technology Network: http://www.oracle.com/technetwork/database/options/clustering/overview/fastapplicationnotification12c-2980342.pdf
See Also:
Oracle Database Administrator’s Guide for more information about using FAN callouts
Oracle Real Application Clusters Administration and Deployment Guide for more information about configuring Fast Application Notification and FAN callouts
Transaction Guard provides a generic tool for applications to use for at-most-once execution in case of planned and unplanned outages and repeated submissions.
Starting with Oracle Database 12c, applications can use a logical transaction ID (LTXID) to determine the outcome of the last transaction open in a database session following an outage. Without using Transaction Guard, applications that attempt to retry operations following outages can cause logical corruption by committing duplicate transactions.
Without Transaction Guard, one of the fundamental problems for recovering applications after an outage is that the commit message that is sent back to the client is not durable. If there is a break between the client and the server, the client sees an error message indicating that the communication failed (also called a recoverable error). This error does not inform the application whether the submission executed any commit operations, or if a procedural call ran to completion while executing all expected commits. The error also does not indicate session state changes or intermittent failures. Worse yet, the application may still be running disconnected from the client.
These recoverable errors can cause end users or applications that have their own replay (instead of using Application Continuity) to issue duplicate transaction submissions, re-issue changes that are already committed in the database, and other forms of "logical corruption." The transaction cannot be validly resubmitted if the non-transactional state is incorrect or if it is committed. Continuing to process a committed but not completed call can result in the application using a database session that is in the wrong state.
In Oracle Database 12c Release 1 (12.1), Transaction Guard provides new, integrated tools for applications to use to achieve idempotence automatically and transparently, and in a manner that scales. Transaction Guard uses the logical transaction ID (LTXID) to eliminate duplicate transactions, a process referred to as transaction idempotence. The logical transaction id persists on commit and is reused following a rollback. During normal runtime, a LTXID is automatically held in the session at both the client and server for each database transaction. At commit, the LTXID persists as part of committing the transaction and the database returns the next LTXID to use to the client.
Example 6-1 How Application Continuity or an Application Uses Transaction Guard
The Oracle Client driver receives a FAN down event.
FAN automatically aborts the terminated session.
If it is a recoverable error, then:
Get the last LTXID from the failed session using the APIs provided by the client driver: getLogicalTransactionId()
for JDBC, LogicalTransactionId
for ODP.NET, and OCI_ATTR_GET
for OCI.
Obtain a new session.
Call the PL/SQL procedure DBMS_APP_CONT.GET_LTXID_OUTCOME
with the last LTXID gotten from the failed session.
Note:
You must grant the EXECUTE
privilege on the DBMS_APP_CONT
package to the database users that use the package.
If the outcome of this transaction is COMMITTED and COMPLETED, then return the result of that transaction to the application.
If the outcome of the session state (such as messages or the PL/SQL block) is COMMITTED, but not COMPLETED, then return the result of that transaction to the application and a warning that the transaction was committed, but there may be additional states such as row counts preventing the application from continuing.
Note:
Most applications will not need these states. This is a consideration for those that do. For those that do not care about row counts or other states from the call, then return COMMITTED to the application.
If the outcome of the transaction is not COMMITTED, then, the user or application can chose to resubmit the last request.
See Also:
Oracle Database Development Guide for more information about Transaction Guard and idempotence
Oracle Real Application Clusters Administration and Deployment Guide for more information about Transaction Guard and client failover for Oracle RAC
Oracle Database JDBC Developer's Guide about using Transaction Guard with Java applications
Oracle Data Provider for .NET Developer's Guide for Microsoft Windows for information about Transaction Guard support for ODP.NET applications
Application Continuity can mask database outages for both planned and unplanned events.
Following planned and unplanned outages, Application Continuity attempts to mask the outage by rebuilding the database session, and resubmitting the pending work following recoverable errors that make the database session unavailable.
When Application Continuity is configured, an end-user request is replayed if it did not complete (if replay is enabled when the outage occurred); replay is started if the time has not exceeded the replay timeout attribute specified for the service. When a component fails or becomes unresponsive, Application Continuity attempts to restore the database sessions back to the current time. When replay succeeds, this feature masks applications from transient outages (such as session failure, instance or node outages, network failures and so on) and from planned outages such as repairs, configuration changes and patching the software.
Before you can use Application Continuity, you must complete various checks.
Before you can use Application Continuity, you must configure attributes for the database services used by the applications. The application may also need to be modified to borrow and return connections rather than pin connections. This is normally a property. Request boundaries need to be added if one of the integrated pools (UCP, WLS data source) is not being used. You may also need to mark request boundaries if not returning connections to the Oracle Connection Pool and a property is not available to unpin. With Oracle Database 12c Release 1 (12.1), Application Continuity for Java is available for general use with the JDBC-Thin Oracle driver, the JDBC Universal Connection Pool, and WebLogic Server Active Grid Link. It is also available with OCI 12c when using PeopleSoft Applications.
To use Application Continuity with an Oracle RAC database, use the following configuration checklist:
Service Configuration Checks
Create a dynamic database service (also referred to as an application service) to connect to the database. Do not use the Oracle SID, instance name, or the default database service to connect to an Oracle RAC or Oracle Data Guard database.
For the service, set failovertype
to TRANSACTION
, commit_outcome
to TRUE
and notification
to TRUE
. Optionally, to find the best connections to use, set rlb_goa
to SERVICE_TIME
and clb_gloal
to SHORT
. See "Creating Services Using SRVCTL".
Software Configuration Checks (Database and Middle-tier)
Use Oracle Database 12c Release 1 (12.1) or later.
Use the JDBC Universal Connection Pool (12.1) or WebLogic Active GridLink (12.1.2 or later) configured with the JDBC Replay data source. You can also use the JDBC Replay data source with your own JDBC connection pool.
If a statement cache at the application server level is enabled (for example, the WebLogic or third-party application server statement cache), then this must be disabled when the replay is used. Instead, configure the JDBC statement cache, which supports Application Continuity. The JDBC statement cache also performs better because it is optimized for JDBC and Oracle. Use oracle.jdbc.implicitstatementcachesize=
nnn
.
Use FAN and FCF for WebLogic Active GridLink Data Source, Universal Connection Pool, or, as an option, use simple JDBC FAN with third party pool.
Check the resource requirements; ensure there is sufficient CPU and memory at the middle tier.
Application Continuity is managed on the server-side, and uses hardware when available for CRCs. On the client side, additional CPU costs are incurred for garbage collection. The replay driver requires more memory than the base driver because the calls are retained until the end of a request. At the end of the request, the calls are released to the garbage collector. This action differs from the base driver that releases closed calls.
Note:
CPU overhead is reduced on the database side for platforms with current Intel and Sparc chips where validation is assisted by the hardware.
Determine whether the application can use mutables, or sequential data that might be different on a different instance. If so, then you configure the application to keep the original values for SYSDATE
, SYSTIMESTAMP
, SYS_GUID
, and sequences during failover.
See Oracle Database SQL Language Reference for information on issuing the GRANT
privilege to users for KEEP
[DATE TIME|SYSGUID|
Sequence
]
.
Grant EXECUTE
on the DBMS_APP_CONT PL/SQL package to the users that will use Application Continuity.
After consulting the application certification, assign the GRANT
privilege for keeping mutables to the users that will use replay.
Check with the application developers to determine if any requests should not be replayed. The application must explicitly call an API to disable replay for those requests.
Application Code Checks (consult with the application developer)
Determine whether the application uses Oracle JDBC concrete classes. If it does, and if access is needed to Oracle proprietary APIs, then plan to replace these concrete classes with standard JDBC or Oracle JDBC interfaces. Refer to My Oracle Support Note 1364193.1 at :
https://support.oracle.com/CSP/main/article?cmd=show&type=NOT&id=1364193.1
Use a connection string that sets connection retries and a delay between these retries. For an example of configuring the connection properties using JDBC, see "Configuring JDBC Clients for Fast Connection Failover".
Decide whether an optional callback will be used for initializing connections from the application to the database. When using Oracle WebLogic Server or the Universal Connection Pool, connection labeling is recommended. If registered, a callback will execute at runtime and replay.
Determine whether replay must be explicitly disabled for any code path in the application. In particular, review external PL/SQL calls such as messaging and callouts and server side Java. If these should not be replayed, then use the disableReplay
API.
Determine whether the application must have begin and end request API's added to the application's own connection pool to identify request boundaries or if the application borrows and returns connections from the WebLogic Server Pool or Universal Connection Pool for each request.
If the application borrows and returns connections from the WebLogic Server Pool or Universal Connection Pool for each request, then no change is needed.
If the application uses Oracle connection pools, and does not return connections between requests, see if there is a property to set to return connections to the pool. If there is no property to set, or if using the application's own connection pool, then you must add beginRequest
and endRequest
boundaries.
See Also:
Oracle Database Development Guide for more information on how Application Continuity works, and how you can use it in applications
Oracle Database Concepts for more information about transactions
Oracle Database JDBC Developer's Guide for more information about Application Continuity for Java applications
Application Continuity is recommended for planned outages combined with draining requests from Oracle connection pools.
Using any FAN-aware pool with FAN configured (OCI, UCP, DRCP, WebLogic Server, or ODP.NET), the steps are as follows:
Use the SRVCTL relocate
command to relocate the service from the instance being shut down without disrupting the sessions (do not use the -force
parameter). If relocating a uniform service, then use srvctl stop service
at the instance (do not use the -force
parameter).
The FAN event for the planned outage clears the idle sessions immediately and marks the active sessions to be released at check in (end of request). This drains the sessions from the instance without disrupting work.
All sessions should check in, but in reality this does not always happen. When sufficient time has elapsed for the sessions to check in, stop the instance using the ABORT
shutdown option. For Application Continuity enabled pools (UCP and WebLogic Server), Application Continuity attempts to recover those remaining sessions that did not check in.
This step applies for UCP and WLS, and any pool that adds beginRequest
/endRequest
on JDBC Thin Replay Driver.
Restart the instance and service. Run-time load balancing, if enabled, balances the sessions back to use the restored instance at the next request boundaries.
The Load Balancing Advisory provides information to applications or clients about the current service levels that the Oracle RAC database instances are providing.
Applications can take advantage of the load balancing Fast Application Notification (FAN) events to direct work requests to the instance in the cluster that provides the best performance based on the workload management directives that you have defined for that service. Also, when an instance is restarted, Oracle RAC uses FAN events to notify the application's connection pool so that the connection pool can create connections to the recently started instance and take advantage of the additional resources that this instance provides.
The load balancing advisory is integrated with the Automatic Workload Repository built into Oracle Database 12c. The Automatic Workload Repository measures response time and CPU consumption for each service.
The advice given by the Load Balancing Advisory takes into account the power of the server and the current workload of the service on the server. Enabling the Load Balancing Advisory helps improve the throughput of applications by not sending work to instances that are overworked, running slowly, not responding, or have failed.
Your application can take advantage of the Load Balancing Advisory without any programmatic changes if you use an integrated Oracle client, one that has the Run-time Connection Load Balancing feature. Due to the integration with FAN, Oracle integrated clients are more aware of the current status of an Oracle cluster. This prevents client connections from waiting or trying to connect to an instance that is no longer available. The integrated clients for FAN events include the Universal Connection Pool (UCP) used by Oracle Database 12c JDBC, Oracle Database 12c ODP.NET, and Oracle Database 12c Oracle Call Interface (OCI).
You configure your Oracle RAC environment to use the Load Balancing Advisory by defining service-level goals for each service used. Defining a service-level goal enables the Load Balancing Advisory for that service and enables the publication of FAN load balancing events. There are two types of service-level goals for Run-time Connection Load Balancing:
Service Time. The Load Balancing Advisory attempts to direct work requests to instances according to their response time. Load Balancing Advisory data is based on the elapsed time for work done by connections using the service, and the available bandwidth to the service. This goal is best suited for workloads that require varying lengths of time to complete, for example, an internet shopping system.
Throughput. The Load Balancing Advisory measures the percentage of the total response time that the CPU consumes for the service. This measures the efficiency of an instance, rather than the response time. This goal is best suited for workloads where each work request completes in a similar amount of time, for example, a trading system.
If you do not select the Enable Load Balancing Advisory option, then the service-level goal is set to None, which disables load balancing for that service.
See Also:
"About Continuous Service Availability with Database Services"
Oracle Real Application Clusters Administration and Deployment Guide for more information about integrated Oracle clients
Oracle Net connection load balancing spreads user connections across all the instances that support the service used to connect to the database.
Oracle Net is a software component that resides on the client and on the Oracle database server. It establishes and maintains the connection between the client application and the server, and exchanges messages between them using industry standard protocols. For the client application and a database to communicate, the client application must specify location details for the database it wants to connect to, and the database must provide some sort of identification or address.
On the database server, the Oracle Net Listener, commonly known as the listener, is a process that listens for client connection requests. The configuration file for the listener is the listener.ora
.
Oracle Database 12c database clients use SCAN and the easy connect method to connect to the database. SCAN can resolve to multiple IP addresses, reflecting multiple listeners in the cluster handling public client connections. When using the easy connect method, you do not have to configure any client network files. You simply specify a connect identifier with the following format:
SCAN[:port]/service_name
SCAN represents the SCAN for your cluster. If you do not specify a port number, then the default value of 1521 is used for the TCP port identifier. The service_name is the name of a dynamic database service.
You can also use Net Configuration Assistant (NETCA) to create a net service name. The net service name resolves to the connect descriptor. The address portion of the connect descriptor is actually the protocol address of the listener. The client uses a connect descriptor to specify the database or instance to which the client wants to connect.
When a net service name is used, establishing a connection to a database instance takes place by first mapping the net service name to the connect descriptor. This mapped information is stored in one or more repositories of information that are accessed using naming methods. The most commonly used naming method is Local Naming, where the net service names and their connect descriptors are stored in a localized configuration file named tnsnames.ora
.
When the client connects to the cluster database using a service, you can use the Oracle Net connection load balancing feature to spread user connections across all the instances that are supporting that service. There are two types of load balancing that you can implement: client-side and server-side load balancing. In an Oracle RAC database, client connections should use both types of connection load balancing. When you create an Oracle RAC database using Oracle Database Configuration Assistant (DBCA), DBCA configures and enables server-side load balancing by default.
See Also:
Oracle Database 2 Day DBA for more information about network configurations
Client-side load balancing balances the connection requests across the listeners.
When the listener receives the connection request, the listener connects the user to an instance that the listener knows provides the requested service.
Client-side load balancing is defined in your client connection definition by setting the parameter LOAD_BALANCE=yes
in the tnsnames.ora
file. When you set this parameter to yes
, the Oracle client randomly selects an address from the address list, and connects to that node's listener. This balances client connections across the available listeners in the cluster.
When you create an Oracle RAC database using DBCA, the assistant creates a sample load balancing connection definition in the tnsnames.ora
file.
Client-side load balancing includes connection failover. With connection failover, if an error is returned from the chosen address, then Oracle Net Services tries the next address in the address list until either a successful connection is made or it has exhausted all the addresses in the list.
See Also:
Oracle Database 2 Day DBA for more information about network configurations
Oracle Database 2 Day DBA for more information about client connections
Oracle Database 2 Day DBA for more information about the tools you can use for configuring the network
With server-side load balancing, the listener directs a connection request to the best instance currently providing the service by using information from the Load Balancing Advisory.
For each service, you can define the method the listener uses for load balancing by setting the connection load balancing goal. You can use a goal of either long or short for connection load balancing. These goals have the following characteristics:
Short—Connections are distributed across instances based on the best response time. Use the Short connection load balancing goal for applications that have connections of brief duration.
Long—Connections are distributed across instances based on the number of sessions in each instance, for each instance that supports the service. Use the Long connection load balancing goal for applications that have connections of long duration. This is typical for connection pools and SQL*Forms sessions. Long is the default connection load balancing goal.
Note:
If you did not use DBCA to create your database, or if you are using listener ports other than the default of 1521, then you must configure the LOCAL_LISTENER
and REMOTE_LISTENER
database initialization parameters for your cluster database to point to SCAN:port
.
Run-time Connection Load Balancing is a feature of Oracle connection pools that can distribute client work requests across the instances in an Oracle RAC database based on the Load Balancing Advisory information.
The connection allocation is based on the current performance level provided by the database instances as indicated by the Load Balancing Advisory FAN events. This provides load balancing at the transaction level, instead of load balancing at the time of the initial database connection.
With Run-time Connection Load Balancing, applications use Load Balancing Advisory information to provide better performance to users. OCI Session pools and ODP.NET connection pools support Run-time Connection Load Balancing. For Java applications, Oracle recommends the Universal Connection Pool (UCP). The Universal Connection Pool is integrated to take advantage of Load Balancing Advisory information. UCP, introduced in Oracle Database 11g patch set 1 (11.1.0.7), can be used against Oracle Database 10g, Oracle Database 11g, or Oracle Database 12c.
You must enable the client data source for Run-time Connection Load Balancing with a service that has the following configuration:
The Load Balancing Advisory is enabled and the service-level goal is set to SERVICE_TIME
, THROUGHPUT
, or NONE
.
The service connection load balancing goal is set to Short.
The illustration in the steps below illustrates Run-time Connection Load Balancing. In this illustration, the Oracle RAC database has three instances. Suppose that the Load Balancing Advisory indicates that Instance1 and Instance3 have the best performance, while Instance2 currently has less than optimal performance. When Run-time Connection Load Balancing is enabled for the Universal Connection Pool, the following process occurs:
A client requests a connection from the connection pool.
Run-time Connection Load Balancing selects the connection that belongs to the most efficient (best) instance from the connection pool. In the illustration in the next step, there are three possible nodes to which the connection can be routed. Instance1, which has the least amount of CPU workload, is currently being assigned about 60 percent of the incoming connections. Instance2, which is currently overloaded, is only being assigned around 10 percent of the incoming connections. Instance3, which has a high workload, is being assigned around 30 percent of the incoming connections. The best instance to handle the connection request in this case would be Instance1.
The client receives the connection that would process the work request with the best response time.
Figure 6-1 Run-time Connection Load Balancing
Starting with Oracle Database 11g you can use an additional flag in the load balancing advisory event called affinity hint. The affinity hint is a flag that indicates if affinity is a benefit for a particular instance and service combination. Different instances offering the same service can have different settings for the affinity hint.
The affinity hint is automatic when the load balancing advisory is turned on through setting the goal on the service. This flag is for temporary affinity that lasts for the duration of a web session. Web conversations often connect and disconnect many times during the entire session. During each of these connects, it may access the same or similar data, for example, a shopping cart, Siebel, and so on. Affinity can improve buffer cache efficiency, which lowers CPU usage and transaction latency.
Applications using Oracle Database 12c and UCP can take advantage of this new affinity feature. If the affinity flag is turned on in the Load Balancing Advisory event, then UCP creates an Affinity Context for the web session such that when that session does a get connection from the pool, the pool always tries to give it a connection to the instance it connected to the first time it acquired a session. The choice of instance for the first connection is based on the current load balancing advisory information.
You can create a service using Oracle Enterprise Manager or the SRVCTL utility.
Note:
TheDBMS_SERVICE
package will create the database structures necessary to offer a service, but none of the Oracle Clusterware functionality is configured, such as service placement, failure characteristics and so on.To manage workloads, you can define database services that you assign to a particular application or to a subset of an application's operations. You can also use services to manage the workload for different types of work.
You can use either Oracle Enterprise Manager Cloud Control or Oracle Enterprise Manager Database Express to create services.
To create a service using Enterprise Manager:
See Also:
Oracle Database Administrator's Guide for more information about the Oracle Scheduler
To configure services for Application Continuity or Transaction Guard using SRVCTL:
You can create and administer services using Enterprise Manager. You can also use the SRVCTL utility to perform most service management tasks.
lsnrctl
utility you can verify that the new service is recognized by Oracle Net and available to the database clients.See Also:
The Cluster Managed Database Services page is the master page for beginning all tasks related to services.
To access this page, go to the Cluster Database Maintenance page, then click Cluster Managed Database Services in the Services section. You can use this page and links from this page to do the following:
View a list of services for the cluster.
View the instances on which each service is currently running.
View the status for each service.
Create or edit a service.
Start or stop a service.
Enable or disable a service.
Perform instance-level tasks for a service.
Delete a service.
See Also:
When managing services using Enterprise Manager, you use the Cluster Managed Database Services page.
To access the Cluster Managed Database Services page:
By configuring and using a SCAN for your cluster, you no longer have to modify the network setup on each client. Using the lsnrctl
utility you can verify that the new service is recognized by Oracle Net and available to the database clients.
Note:
When using utilities to manage your cluster, databases, database instances, Oracle Automatic Storage Management (Oracle ASM), and listeners, use the appropriate binary that is in the home directory of the object or component you are managing. Also, set your ORACLE_HOME
environment variable to point to this directory.
For example, to use lsnrctl
to manage the SCAN listener, use the binaries located in the Grid home (where the listener is running), and set the ORACLE_HOME
environment variable to the location of the Grid home.
To verify Oracle Net Services supports the newly created service:
See Also:
Oracle Database 2 Day DBA for details on how to view the listener configuration
Oracle Database 2 Day DBA for details on how to connect to an Oracle database from a client computer
There are three central elements to consider when automating failover for application clients.
First, clients that are connected at the time of failure must be quickly and automatically notified that a failure has occurred to avoid waiting for TCP/IP network time-outs before attempting to connect to the new database instance (such time-outs range anywhere from eight minutes to two hours, depending on operating system). Oracle RAC configurations use Fast Application Notification (FAN) to notify JDBC clients, OCI clients, and ODP.NET clients. FAN event notifications and callouts enable automatic and fast redirection of clients after a site failure.
The second central element of client failover, is the redirection of new incoming requests to the new instance after a failure has occurred, which can be implemented using application services. When you create and activate application services on several instances of an Oracle RAC database, then if the service becomes unavailable on one instance, clients connections are redirected to other instances where the service is available. Applications need only provide the name of the database service in their connect strings–there is no need to specify an instance name because, using listener cross registration, SCAN listeners and other remote listeners in the cluster are aware of which instances are currently providing a service when a connection request comes in.
The third central element is masking the outage from the client and the application. Masking outages of the database session is a complex task for application development and, therefore, errors and timeouts are often exposed to the client. Application Continuity attempts to mask outages from applications by replaying incomplete application requests after unplanned and planned outages.
For applications that are unable to take advantage of Application Continuity, two additional features are available: Transparent Application Failover (TAF) for OCI based applications and Transaction Guard. Transaction Guard enables applications to build their own masking of outages by knowing the outcome of the last in-flight transaction.
This section deals with configuring high availability for application clients.
See Also:
Oracle Real Application Clusters Administration and Deployment Guide for information about client-side connection configuration
When using services to connect to a database, there are various parameters you can set to configure high availability.
See Also:
Oracle Database Net Services Reference for more information about the timeout and retry parameters for Oracle Net Services
You can configure JDBC clients to use a connection pool, Application Continuity and Transaction Guard to make them highly available.
An application that uses JDBC connections should use a connection pool, so an existing connection can be used in the pool instead of making a new connection to the database every time you access data. The Universal Connection Pool (UCP) is a Java-based connection pool that supports any type of connection (JDBC, LDAP, JCA), to any type of database (Oracle or non-Oracle) with any middle tier (Oracle or non-Oracle). It also supports standalone deployments such as TopLink or BPEL. UCP also supports integration features of Oracle Database such as Fast Connection Failover, Run-time Connection Load Balancing, Web Affinity, Application Continuity, and Transaction Guard with Oracle RAC when using the JDBC Thin driver.
Without Fast Connection Failover, if a node or network goes down and the application attempts to connect to that instance, then the connection request might hang for several minutes waiting for the TCP time out to be reached. If you chose not to use a connection pool with the integrated FCF feature, then your application can still be notified of Oracle RAC high availability events by configuring the Connection Failure Notification feature.
Application Continuity provides a general purpose, application-independent infrastructure that enables recovery of work from an application perspective, after the occurrence of a planned or unplanned outage. The outage can be related to the system, communication, or hardware layer following a repair, a configuration change, or a patch application. Application Continuity restores the database session by replaying the non-transactional and transactional user calls to reconstruct the database session as if the outage had not occurred. Application Continuity performs this replay beneath the application so that the outage appears to the application as a delayed execution.
Transaction Guard is a reliable protocol and tool that returns the outcome of the last in-flight transaction after an outage makes the database session unavailable. Transaction Guard avoids the costs of ambiguous errors that lead to end user frustration, customer support calls, and lost opportunities.
Note:
Transaction Guard does not replay distributed or XA transactions.
When using the Universal Connection Pool (UCP), configure Fast Application Notification and, optionally, Application Continuity. Do not configure Connection Failure Notification (the ConnectionFailureNotification
connection property).
ConnectionFailureNotification
connection property.oracle.jdbc.replay.OracleDataSource
) serves as the front-end for applications to obtain a connection when using Application Continuity for Java.See Also:
Oracle Real Application Clusters Administration and Deployment Guide for more information about deploying Oracle RAC for high availability
Oracle Universal Connection Pool for JDBC Developer's Guide for more information about the Universal Connection Pool for Java
Follow these steps to configure JDBC Clients for Fast Connection Failover.
The Universal Connection Pool (UCP) subscribes to FAN when Oracle Notification Service is configured and the pool property setFastConnectionFailoverEnabled
is enabled.
UCP also subscribes to the FAN Load Balancing events if you configure the dynamic database service used in the connection for run-time load balancing. Instead of randomly assigning a free connection to a work request, the connection pool chooses the connection that gives the best service according to the latest information it has received. If a node becomes hung, then run-time load balancing quickly shifts connections from the hung node to other nodes in the cluster.
To configure JDBC clients for Fast Connection Failover:
See Also:
Oracle Database JDBC Developer's Guide for more information about installing and configuring Oracle RAC Fast Application Notification for Java
Oracle Universal Connection Pool for JDBC Developer's Guide for more information about using the Universal Connection Pool for database connections using Java
Oracle Database 2 Day + Java Developer's Guide for information about creating a method to authenticate users
Oracle Real Application Clusters Administration and Deployment Guide for information about installing and configuring Oracle RAC Fast Application Notification
To configure simple FAN for non-pooled JDBC clients, you use the ConnectionFailureNotification
connection property.
JDBC simple FAN enables standalone Oracle JDBC connections to quickly respond to node failures. This feature enables a standalone JDBC Thin Driver connection to listen for FAN events and respond to a node down event by marking the connection closed, permitting user applications to begin recovery quicker. It does not alter the behavior of the application, except that JDBC method calls throw exceptions more quickly after a node goes down.
To configure FAN for non-pooled JDBC Clients:
See Also:
Oracle Database JDBC Developer's Guide for information about installing and configuring Oracle RAC Fast Application Notification
Oracle Database 2 Day + Java Developer's Guide for information about creating a method to authenticate users
A new Replay data source (oracle.jdbc.replay.OracleDataSource
) serves as the front-end for applications to obtain a connection when using Application Continuity for Java.
JDBC applications using Oracle JDBC Replay data source and using neither UCP or WebLogic Server Active GridLink—typical standalone JDBC cases
JDBC applications using UCP data sources—standalone or 3rd party application servers configured to use a UCP data source
JDBC applications using only WebLogic Server Active GridLink but not UCP data sources—typical WebLogic Server J2EE cases
To configure JDBC clients to use the JDBC Replay Driver:
Note:
If you disconnect or terminate a session that uses Application Continuity, then Application Continuity attempts to recover the session. To disconnect or terminate the session without replay, see Oracle Database Development Guide.
See Also:
Oracle Database JDBC Developer’s Guide for information about configuring Oracle JDBC for Application Continuity for Java
Oracle Database Development Guide for more information about using Application Continuity
Oracle Universal Connection Pool Developer’s Guide for information about using Application Continuity with the Universal Connection Pool
Transaction Guard is a reliable protocol and tool that returns the outcome of the last in-flight transaction after an outage that makes the database session unavailable.
Transaction Guard avoids the costs of ambiguous errors that lead to user frustration, customer support calls, and lost opportunities. Without Transaction Guard, applications and users who attempt to retry operations following an outage can cause logical corruption by committing duplicate transactions or committing transactions out of order.
See Also:
Oracle Database JDBC Developer's Guide for information about configuring JDBC-Thin clients for use with Transaction Guard
Oracle Database Development Guide for more information about using Transaction Guard
The Oracle Call Interface (OCI) provides integration with FAN and Load Balancing Advisory events. To take advantage of the Load Balancing Advisory, you must enable the OCI Session pool.
See Also:
"Using Transaction Guard" in Oracle Database Development Guide
"Configuring Transparent Application Failover" in Oracle Call Interface Programmer's Guide
Oracle Call Interface (OCI) clients can register to receive notifications about Oracle Real Application Clusters (Oracle RAC) high availability events and respond when events occur.
Registering to receive Fast Application Notification (FAN) events improves the connection failover response time in OCI and also removes terminated connections from connection and session pools. This feature works for all OCI client applications.
To configure OCI clients to receive FAN notifications:
See Also:
Oracle Call Interface Programmer's Guide for more information about event notification and user-registered callbacks
Oracle Real Application Clusters Administration and Deployment Guide for more information about configuring fast application notification for OCI clients.
Oracle Call Interface (OCI) clients can use Transaction Guard to replay an in-flight transaction when a recoverable error occurs.
OCI supports Fast Application Notification (FAN) messages and Transparent Application Failover (TAF). FAN is designed to quickly notify an OCI-based application of outages at the node, database, instance, service, and public network levels. Once notified of the failure, an application can leverage TAF to re-establish the failed connection on a surviving instance. What is missing is the ability to reliably determine the outcome of the failed transaction and the ability to recover the current transaction after database connectivity has been restored. This section describes how to configure OCI to enable transactional TAF. Transactional TAF is the OCI driver's ability to replay a transaction that is active when a recoverable error occurs.
To configure OCI clients to use Transaction Guard:
See Also:
Oracle Call Interface Programmer's Guide for more information about event notification and user-registered callbacks
Oracle Real Application Clusters Administration and Deployment Guide for more information about configuring fast application notification for OCI clients.
Oracle Data Provider for .NET (ODP.NET) provides integration with Fast Application Notification (FAN), Run-time Connection Load Balancing, Transparent Application Failover (TAF), and Transaction Guard.
When a connection to an Oracle instance is unexpectedly severed, TAF seamlessly attempts to failover to another Oracle instance. Due to the delay that a failover can incur, the application may wish to be notified by a TAF callback. ODP.NET supports TAF callback through the Failover event of the OracleConnection
object. To receive TAF callbacks, an event handler function must be registered with the Failover
event of the OracleConnection
object. Also the connection parameter enlist
should be set to false
for TAF to work.
ODP.NET connection pools subscribe to FAN notifications from Oracle RAC that indicate when nodes are down and when services are up or down. Based on these notifications, the ODP.NET connection pool:
Interrupts active sessions so they do not wait on TCP/IP timeouts, and
Clears idle sessions previously connected to the down node or instance or network so they are not passed to user sessions
Creates new connections to healthy nodes, if possible
ODP.NET provides Run-time Connection Load Balancing to provide enhanced load balancing of the application workload. Instead of randomly selecting an available connection from the connection pool, it chooses the connection that can provide the best service based on the current workload information.
The procedures for enabling ODP.NET are similar to the procedures for enabling JDBC in that you must set parameters in the connection string to enable Fast Connection Failover (FCF).
To configure ODP.NET clients to receive FAN notifications:
See Also:
Oracle Data Provider for .NET Developer's Guide for Microsoft Windows for more information about event notification and user-registered callbacks
Oracle Real Application Clusters Administration and Deployment Guide for more information about configuring fast application notification for ODP.NET clients.
When an ODP.NET application submits a transaction commit request and some type of failure occurs, such as a node, network, or database failure, ODP.NET can now deterministically conclude whether the transaction committed or not.
Starting with Oracle Database 12c, applications can use a logical transaction ID (LTXID) to determine the outcome of the last transaction open in a database session following an outage. Using Transaction Guard, the end user experience is vastly improved by returning to the application and user, following an outage, whether the last submission committed and completed or did not. A committed result stays committed. An uncommitted result stays uncommitted, and is a green light, for example, for the user or application to resubmit.
Without using Transaction Guard, applications that attempt to retry operations following outages can cause logical corruption by committing duplicate transactions.
See Also:
Oracle Data Provider for .NET Developer's Guide for Microsoft Windows for an example of using Transaction Guard in an ODP.NET application
Oracle Real Application Clusters Administration and Deployment Guide for more information about configuring fast application notification for ODP.NET clients.