Container server zones
Container servers have zones that are used to distribute data replicas.
Zones are logical groupings of physical resources, such as a server, a rack of servers, floors of a building, and data centers. When you define a container server, you provide the zone name for that container. The zones protect against data loss in case your physical resources fail.
Data replicas are not stored in the same zone as the primary data. If you want data to be replicated, you must name at least two different zones.
For example, if your system is set up with three physical servers, each server contains all of the containers for one zone (Container_1 zone_A, Container_2 zone_B, and Container_3 zone_C). Two data replicas are automatically created and stored in the following manner:
| Primary data in ... | Replicated data in ... |
|---|---|
| Container_1 zone_A |
|
| Container_2 zone_B |
|
| Container_3 zone_C |
|
In this situation, if one or even two servers fail, data is not lost because the data is replicated to the other two servers. A maximum of two replicas are created. If you create six container zones, only two replicas of any data set are created, not five replicas.
If you have container servers that are likely to fail together, place those servers in the same zone. These servers might be container servers that are running on the same server or rack.