High availability recommendations

components are virtual machines (VMs) that run on VMware ESX/ESXi hosts. They can, therefore, use VMware High Availability (VMware HA), a feature of the VMware vSphere virtualization platform, which restarts a VM if it becomes unresponsive on its host or on another host if the original host is also unresponsive.

This topic provides recommendations for providing high availability (HA) for

components.

Requirements

VMware HA is available for

components, provided that you satisfy the following technical requirements:

• Hosts must use the Domain Name System (DNS).
• Hosts must support reverse lookup.
• Target hosts must provide connectivity to the same networks, the same network label (in other words, the same port-group name), and the same IP subnet, VLAN, and physical network access.
• Data must be stored on a shared storage device, and target hosts must have access to this data.
• If Distributed Resource Scheduler (DRS) is enabled, it must not be in manual mode.
• If Distributed Power Management (DPM) is enabled, it must not be in manual mode.

Recommendations

BMC recommends the following strategies for providing HA for

components.

General

To increase the likelihood that a VM has sufficient resources when it restarts, set its HA restart priority to High.

The macro unmigrated-inline-wiki-markup from Confluence is no longer available.

For all hosts that are fail over targets for Collector instances, you must configure the capture port network identically. This requirement might necessitate additional taps or span ports. If not, the VM might failover but not capture any traffic data. BMC recommends that you test all failover targets to ensure that traffic feeds are available to them. You can also use VM-Host affinity must rules to bind a Collector VM to a particular host group.

The macro unmigrated-inline-wiki-markup from Confluence is no longer available.

Do not configure the isolation response to switch power off. Also, BMC recommends that you test all failover targets to ensure that Collector feeds are available to them.

The macro unmigrated-inline-wiki-markup from Confluence is no longer available.

Do not configure the isolation response to switch power off. Also, to avoid data loss, BMC recommends that you configure the off-board storage device to terminate write operations cleanly.

Additional resources for high-availability information

The following sites, which are not created or endorsed by BMC, provide information that you might find helpful:

• vSphere Availability Guide (Version 4.1)

• VMware High Availability (VMware HA): Deployment Best Practices

• VMware vSphere 4.1 Documentation

• VMware vSphere 4 - ESXi Installable and vCenter Server Documentation Center