Why do I need to define a cluster name for Swarm?

Defining a cluster name is a best practice to meet numerous requirements and support the fullest use of Swarm.

Centralized DR clusters — Using defined cluster names is required to support centralized DR clusters using wide area replication. Cluster names allow the central DR cluster to track content from multiple clusters and replicate back to the appropriate source clusters. You should give each cluster a fully qualified domain name that easily identifies it, such as hq.caringo.com.

Tenanting unnamed objects — The default cluster domain is an optional domain whose name exactly matches the cluster name. Every unnamed object and every named object that has no domain explicitly defined for it belongs to the default cluster domain (unless you have cluster.enforceTenancy=True disabled, for backwards compatibility). 

Creating a default cluster domain also enables your cluster to use the following features:

• POST authentication for unnamed objects
  
  
• Disambiguation of cluster names in a roll-up cluster
  • The domain name is stored as part of the object's system metadata. As long as every domain name that you manage is unique, there will never be a collision between named objects from different clusters if those clusters are combined (such as in a disaster recovery cluster).

• Simpler object operations
  
  • Application developers do not have to explicitly set a domain name as the Host in the request to perform operations on an object (such as create, update, or delete).