Discovery
Discovery is the first step when a VMware Intelligent Policy (VIP) is used. This step is applicable only when automatic selection of virtual machines for backup via a VIP is enabled. It discovers virtual machines and filters them using the selection query specified in the policy. The resulting list determines which virtual machines are backed up.
To determine which virtual machines need to be protected, data for all virtual machines for each virtual server (vCenter / ESX) is fetched and evaluated. This is usually a fast operation. Data, once queried, is cached on the discovery host and reused. Typically, data that has been fetched within one hour is reused, but there are circumstances in which new data is fetched. You can control which discovery host is used for each policy; this can have a significant impact on backup and virtual infrastructure.
Consolidated discovery hosts
In some cases, it might be beneficial to use a fewer number of discovery hosts across policies. Each discovery host will query for all possible virtual machines from every configured virtual machine server by reusing discovery hosts from another policy. Subsequent policies can benefit from the cache built from a previous discovery. Reducing the number of times discovery is done can also reduce the load on the virtual infrastructure. This is especially a problem if various policies with multiple storage units (media servers) share the same backup window.
Consolidated policies
Policies are useful to help organize backup lifecycle and provide multiple levels of service as required. For VMware backups, there is one discovery job done for each policy, more number of policies there are more amount of discovery jobs would be running. Consolidating several policies can reduce the number of times vCenters are queried for the data. Careful consideration of all the VMware policies is recommended, especially in the case of a large VMware environment. Another consideration for grouping correct VMs in policies is to have better control over resource utilization across the NetBackup domain.
Advanced policy attribute - VMware Server List
This attribute is a mechanism to control which virtual machine servers that NetBackup communicates with for a given policy. In large virtual environments, you can use this list to improve backup/discovery performance. This mechanism can also be used to temporarily avoid a particular vCenter and ESX server, or to eliminate accessing vCenters that are known to not have VMs to be included in the policy.
DNS Configuration
NetBackup may be unable to identify virtual machines when you use the Browse for Virtual Machines dialog. Virtual machine hostnames may not be properly configured in your Domain Name Server system (DNS), or the DNS system may be slow.
In a large VMware environment, reverse name lookups can be very slow depending on the number of virtual machines being discovered. You can change the VNET_OPTIONS option to determine how many items NetBackup can cache.
Another option would be to use VM display name as the Primary VM identifier. This can address problems such as:
Host name/DNS not configured correctly
DNS server is slow to respond
However, if using VM display name as the Primary VM identifier, it is recommended that VM display name is set to same as the VM host name. If Primary VM identifier is different than the host name, while doing individual file and folder restores using the Backup, Archive and Restore (BAR) GUI, backup image selection might not happen automatically. Use the VM search feature to identify correct VM backups to restore from.
Note that each item in NBU's DNS cache takes about 1 KB in memory. This value is in the
bp.conffile on UNIX and Linux and the registry on Windows. See the NetBackup for VMware Administrator's Guide for details.