Explanation/Reference: https://docs.splunk.com/Documentation/Splunk/7.3.2/Indexer/Decommissionasite

Syslog is a standard protocol for sending log messages from various devices and applications to a central server. Syslog can use either UDP or TCP as the transport layer protocol. UDP is faster but less reliable, as it does not guarantee delivery or order of the messages. TCP is slower but more reliable, as it ensures delivery and order of the messages. Therefore, to improve the reliability of syslog delivery to Splunk, it is recommended to use TCP syslog.
Another option to improve the reliability of syslog delivery to Splunk is to use one or more syslog servers to persist data with a Universal Forwarder to send the data to Splunk indexers. This way, the syslog servers can act as a buffer and store the data in case of network or Splunk outages. The Universal Forwarder can then forward the data to Splunk indexers when they are available.
Using a network load balancer to direct syslog traffic to active backend syslog listeners is not a reliable option, as it does not address the possibility of data loss or duplication due to network failures or Splunk outages.
Configuring UDP inputs on each Splunk indexer to receive data directly is also not a reliable option, as it exposes the indexers to the network and increases the risk of data loss or duplication due to UDP limitations.