High availability considerations


TrueSight Capacity Optimization supports levels of high availability (HA). The following table lists the failure characteristics of the components of TrueSight Capacity Optimization:

Component

Single point of failure?

Approach for high availability

Data Warehouse

Yes

Use multiple machines and Oracle RAC option.

Web application

Yes, if only one instance

Use a commercial cluster manager for A/P.

Install on multiple machines, and use load balancer for A/A.

Data hub and Primary scheduler

Yes

Use a commercial cluster manager for A/P.

ETL Engine and Scheduler

Yes, for its portion of data connectors

Use a commercial cluster manager for A/P.

Installation directory

Yes

Use either a storage replication or host-based mirroring.

Content Repository

Yes

Use either a storage replication or host-based mirroring.

For more information, see the following sections: 

High availability of TrueSight Capacity Optimization Data Warehouse via Oracle RAC

The TrueSight Capacity Optimization Data Warehouse is the most critical component of a TrueSight Capacity Optimization deployment. When the Data Warehouse is hosted on a single-node Oracle database, this database is the most important single point of failure.

You can configure the Data Warehouse for high availability (HA) by using Oracle Real Application Clusters (RAC). When you use Oracle RAC for the Data Warehouse, this level of high availability is built into TrueSight Capacity Optimization installation and configuration. During TrueSight Capacity Optimization installation, select High Availability Information, and then type the JDBC URL, which is the URL for Oracle RAC Java database connectivity.

High availability of TrueSight Capacity Optimization machines in various configurations

Apart from the database, the most common configurations of TrueSight Capacity Optimization computers are:

Configuration

TrueSight Capacity Optimization components installed

AS-ALL

Web application, Data hub, Primary scheduler, Analysis Server

AS-WEB

Web application, Analysis Server

AS-HUB

Data hub, Primary scheduler

EE

ETL Engine, scheduler

The following methods enable HA for each of the above configurations.

All of the Active-Passive (A/P) failover configurations below require third-party OS clustering software, for example, Linux Heartbeat or Red Hat Cluster Manager. The following procedures provide general instructions to enable failover to be implemented for TrueSight Capacity Optimization services. The specific scripts and commands will be unique to the third-party package you use.

The Active-Active (A/A) configuration below requires a load balancer. Use this general requirement for configuring TrueSight Capacity Optimization web services behind the load balancer; the specific scripts and commands will be unique to the load balancer you use.

HA for AS-ALL configuration

The purpose of this approach is to configure the product by using the Virtual hostname (also known as Cluster hostname).

In case of failover, the third-party cluster software detects the availability of the nodes and moves the virtual IP, the virtual host name, the storage and the processes to the failover node. You must configure the cluster software to move the resources. An external load balancer might not be able to move the resources.

Configure two machines M1 and M2 in A/P failover by using OS clustering software.

Configure the following resources to be shared and enabled among M1 and M2:

  • A floating IP address associated with the TrueSight Capacity Optimization URL via DNS
  • Front-end web server (Apache), when configured
  • TrueSight Capacity Optimization console JSP engine (Apache Tomcat)
  • Primary scheduler
  • Data hub (Apache Tomcat)
Steps for installing in A/P configuration

You need:

  • Primary machine M1, secondary machine M2
  • Third-party clustering software (Including Virtual IP and Virtual Hostname)
  • Shared or synchronized file system between M1 and M2 for installation files
  • Shared or synchronized directory for TrueSight Capacity Optimization Content Repository

Steps:

  1. Configure clustering software (typically scripts) for failover and giveback (see Failover and giveback procedures ).
  2. Temporarily change the hostname of M1 to the cluster name using hostname command as root.
  3. Install TrueSight Capacity Optimization on M1 and wait for the post-installation process to be completed.
  4. Change the hostname of M1 back to the individual machine name.
  5. Shut down TrueSight Capacity Optimization services.
  6. Move the installation directory into a shared or synchronized mount point, and leave a symbolic link in the original installation directory.
  7. Create a similar symbolic link on machine M2.
  8. Restart TrueSight Capacity Optimization services on M1 and start using TrueSight Capacity Optimization.

Now M1 will be used as primary, and M2 as secondary.

AS-WEB configuration: Web application

For AS-WEB configuration, there are two options:

  1. Configure two machines M1 and M2 in A/P failover.
  2. Configure as many machines as needed in A/A using a load balancer.

Option 1: Configure two machines M1 and M2 in A/P failover:

  • Configure the following resources to be shared and enabled among M1 and M2:
    • A floating IP address associated with the TrueSight Capacity Optimization URL via DNS
    • Front-end web server (Apache), when configured
    • TrueSight Capacity Optimization Console JSP engine (Apache Tomcat)
  • Follow the steps in AS-ALL configuration above.

Option 2: Configure as many machines as needed in A/A by using a load balancer.

A/A configuration for AS-WEB machine

You need:

  • N machines, M1, M2, ..., MN
  • Load balancer, either software or hardware
  • Shared TrueSight Capacity Optimization Content Repository folder between these machines, for example via NFS.
    • For the TrueSight Capacity Optimization Content Repository storage, ensure high availability via storage replication, mirroring, or other means.

Steps:

  1. Install AS-WEB on each machine independently.
    • When installing on the first machine, note down the hostname and the Apache web server consult port number.
    • For the subsequent machines, on the installer's Connect with existing TrueSight Capacity Optimization Server page, select LAN Connection, and then type in the address and Apache web server consult port number. The TrueSight Capacity Optimization installation programs recognize that the additional application server node or nodes are part of the same TrueSight Capacity Optimization deployment as the first node.
    • For each machine, provide the location of the shared Content Repository folder.
  2. Configure load balancer to distribute requests to all N machines.
    • The load balancer must use the hash session policy to preserve connection and session persistence.
  3. Complete the rest of the TrueSight Capacity Optimization installation on AS-HUB and EE machines.

AS-HUB configuration: Data hub, Primary scheduler

The purpose of this approach is to configure the product by using the Virtual hostname (also known as Cluster hostname).

In case of failover, the third-party cluster software detects the availability of the nodes and moves the virtual IP, the virtual host name, the storage and the processes to the failover node. You must configure the cluster software to move the resources. An external load balancer might not be able to move the resources.

Configure two machines M1 and M2 in A/P failover by using OS clustering software.

Configure the following resources to be shared and enabled among M1 and M2:

  • A floating IP address
  • Primary scheduler
  • Data hub (Apache Tomcat)

Follow the steps in AS-ALL configuration above.

EE configuration: ETL Engine, scheduler

Configure two machines M1 and M2 in A/P failover using OS clustering software.

Configure the Scheduler resource to be shared and enabled between M1 and M2.

Follow the steps in AS-ALL configuration above.

Failover and giveback procedures

Use the scripting facilities offered by the third-party clustering software to automate the following procedures. The term script refers to stand for the appropriate scripts in the clustering software.

Automated failover procedure

Triggered by failure of primary (M1) machine

  1. The clustering software must detect that M1 is down and execute the following steps.
  2. On M2, script starts TrueSight Capacity Optimization services with cpit start command.
  3. When M1 restarts, script disables TrueSight Capacity Optimization crontab entries so that TrueSight Capacity Optimization services will not start automatically. This must be done before cron restarts TrueSight Capacity Optimization services.
Automated giveback procedure
  • Applies when TrueSight Capacity Optimization services are running on a secondary node M2.
  • Triggered by clustering software detecting that a primary M1 is available
    1. On M2, script shuts down TrueSight Capacity Optimization services with cpit stop command.
    2. On M1, script starts TrueSight Capacity Optimization services with cpit start command.
Manual giveback procedure
  • Applies when TrueSight Capacity Optimization services are running on a secondary node M2.
  • Triggered by a user command to clustering software (if supported)
  • Or, could be supported by an administrator manually by executing the first step below.
    1. On M2, script shuts down TrueSight Capacity Optimization services with cpit stop command.
    1. On M1, script starts TrueSight Capacity Optimization services with cpit start command.
User initiated failover procedure by clustering sofware command (if supported)
  • Applies when TrueSight Capacity Optimization services are running on a primary node M1.
  • Triggered by a user command on clustering software
    1. On M1, script shuts down TrueSight Capacity Optimization services with cpit stop command
    2. On M2, script starts TrueSight Capacity Optimization services with cpit start command
User initiated failover procedure by manual commands on M1
  • Applies when TrueSight Capacity Optimization services are running on a primary node M1.
  • Triggered by an administrator manually shutting down TrueSight Capacity Optimization services on M1 with cpit stop command
    1. Clustering software must detect that TrueSight Capacity Optimization services are not running on M1
    2. On M2, script starts TrueSight Capacity Optimization services with cpit start command

Migrating a machine running TrueSight Capacity Optimization to A/P configuration

For the four configurations listed above (AS-ALL, AS-WEB, AS-HUB, and EE), it is possible to take a single-machine installation and migrate it to an A/P HA configuration as follows.

Steps for migrating a machine running TrueSight Capacity Optimization to an A/P configuration

You need:

  • Primary machine M1 running TrueSight Capacity Optimization, Secondary machine M2
  • Third-party clustering software ((Including Virtual IP and Virtual Hostname))
  • Shared or synchronized file system between M1 and M2 for installation files
  • Shared or synchronized directory for TrueSight Capacity Optimization Content Repository

Steps:

  1. Configure clustering software (typically scripts) for failover and giveback (see section above).
  2. Temporarily change the hostname of M1 to the cluster name using hostname command as root.
  3. Run setup.sh to reconfigure TrueSight Capacity Optimization.
  4. Change the hostname of M1 back to the individual machine name.
  5. Shut down TrueSight Capacity Optimization services.
  6. Move installation directory into shared or synchronized mount point, and leave a symbolic link in the original installation directory.
  7. Create a similar symbolic link on machine M2.
  8. Restart TrueSight Capacity Optimization services on M1 and start using TrueSight Capacity Optimization.

Now M1 will be used as primary, and M2 as secondary.

HA for AS-ALL configuration
Was this page helpful? Yes No Submitting... Thank you

Comments