Jump to: navigation, search

Resource Manager High Availability

This topic describes how to configure the Genesys Voice Platform (GVP) Resource Manager and the Configuration Database for High Availability (HA) on Windows and Linux operating systems.

Overview

High availability for Resource Manager is achieved by combining two Resource Manager instances in an HA-pair, which enables the resource-management function to be distributed over two servers to provide redundancy. The active Resource Manager node (running on server 1) simultaneously handles the incoming IP traffic and updates the other Resource Manager node (in the HA-pair) with information about the active sessions, thereby achieving high availability and scalability.

HA prevents single point-of-failure and provides immediate failure recovery. In addition, administrators can manage the HA-pair as a single system either locally or remotely.

HA Modes

To configure the Resource Manager Application in HA mode, use the cluster.ha-mode parameter to set the value to active standby, active active, or None. A value of None, which is the default, means the Resource Manager is in stand-alone mode. Two HA modes exist for Resource Manager:

  • Active Standby: Where only the active instance processes SIP requests and the backup instance processes SIP requests when the active instance fails. Two Resource Manager instances share the same virtual IP address but only one instance is actively receiving network traffic.

    To configure the Resource Manager in active standby mode, see Task Summary: Configuring the RM in HA Active Standby Mode (Windows) and Configuring the RM in High Availability Active-Standby Mode (Linux)
  • Active Active: Where either one of the active nodes can process SIP requests. This mode is similar to the virtual IP solution; however, an external load balancer maintains the virtual IP address that forwards information to the active nodes in the cluster. The load balancer can apply proprietary load-balancing rules when it forwards the requests.

    The virtual IP address for the active active HA-pair is configured by using the cluster.virtual-ip parameter in the Resource Manager Application. The Resource Manager advertises the virtual IP address in the Record-Route and Via headers so that SIP Server or the Media Control Platform can send mid-call requests and responses to the load balancer.

    To configure the Resource Manager in active active mode, see Task Summary: Configuring the RM in HA Active-Active Mode (Windows) and Configuring the RM in High Availability Active-Active Mode (Linux)
Tip
When configured as an HA-pair, both Resource Manager instances must manage the same set of resources.

In addition, the load balancer must have the routing capability to route the same SIP transactions to the same Resource Manager instance.

(The load balancer must be SIP-aware and support the Call-ID stickiness feature to ensure this requirement is met.) The same Resource Manager node must process the SIP transaction in its entirety.

Before You Begin

Before you begin the procedures in this section, pay attention to the following information:

Ensure that the Resource Manager hosts...
  ...are configured and fully functional in stand-alone mode.
  ...reside on the same subnet.
  ...have at least two network interface cards (NICs) that are configured with unique IP addresses (within the same subnet).
  ...have one virtual IP address allocated for the cluster (to be shared by the Resource Manager hosts in the HA-pair)

Tip
If you are configuring HA on Linux by using the Simple Virtual IP failover option, only one NIC is required. However, to avoid a single point-of-failure if the NIC or network switch fails, you must use the Bonding Driver failover option with two NICs. The bonding driver is configured to share one IP address among two NICs operating one NIC as active and the other as standby. See Resource Manager HA (Linux). If you configure active active HA on either Windows or Linux, only one NIC per server is required.
  • If you are deploying the Supplementary Services Gateway (SSG) on the same host as the Resource Manager, be aware of a possible port conflict. The default HTTPS Port for the Supplementary Services Gateway is 9801 (controlled by the HTTPS Port parameter in the HTTP section), which is also the default port value for the Resource Manager member.1 and member.2 cluster parameters when they are configured for HA. Use Genesys Administrator to change the values, if required.
  • When you configure the Resource Manager for active standby HA, five IP addresses are required. (If you are using the Simple Virtual IP failover option, only three IP addresses are required.) The first network adapter handles the network traffic that is addressed to the server as part of the HA-pair, while the second network adapter is used for intrahost communication. Figure 32 depicts an example of how IP addresses are assigned to the servers.
  • Simple Network Management Protocol (SNMP) Management Information Bases (MIB) function properly when there is only one instance of a GVP component on a host. It is not recommended that you install the SNMP MIBs on a server that is hosting more than one instance of the Resource Manager. SNMP traps are sent by the active Resource Manager only.
Figure: IP Address Assignation for active standby HA

Monitoring the NICs

When the Resource Manager is configured as an active standby HA-pair, it monitors the NICs to determine network error instances for example, when a cable is unplugged. If any of the NICs encounters errors, the network is considered to be down.

The Resource Manager must monitor the NICs that are used in the HA-pair to ensure that the Network Load Balancing (NLB) is working properly:

  • For Windows: One with NLB configured, and one with the IP address of the host computer.
  • For Linux: If only one NIC is used in the HA-pair, it must be monitored.

When there are more than two NICs present in the system and the other NICs are not part of the NLB HA-pair, it is recommended that you specify the Media Access Control (MAC) addresses of the NICs that you want to monitor.

To specify the NICs that you want to monitor, use the steps in Procedure: Specifying the NICs to Monitor (Windows), and Procedure: Specifying the NICs for Monitoring (Linux).

Tip
When Windows NLB is used, the standby Resource Manager can take up to seven (7) seconds to become active when failover occurs. However, there is no data loss, even for those calls that are in progress.

Integrating GVP with SIP Server for an Active-Active Resource Manager Configuration

For all GVP deployments with SIP Server that will support Resource Manager High Availability Active-Active in IP modes, configure a VoIP service DN.

[+] GVP as a Self-service IVR
[+] GVP as an Inbound Media Server
[+] GVP as an Outbound Media Server
[+] Limitations of the Active-Active RM Configuration

Feedback

Comment on this article:

blog comments powered by Disqus
This page was last modified on July 16, 2018, at 02:43.