Copying and replicating a [NoSQL Distilled] family of column database

In section 4.5, Combination of fragments and replication of a DistSQL book without distillation, the following statement is made:

"Using peer-to-peer replication and outlines is a common strategy for column databases."

The statement does not indicate other types of databases ready for the cluster, namely key stores and document stores. Why is this so? Are these databases well-defined, but not combined with peer-to-peer replication? Is master-slave replication better suited in these cases?

+4
source share
1 answer

Replication between peers is more related to the consistency model. You compromise between fault tolerance and consistency when the peer-to-peer model selects the former and the master-slave model of the latter. Consistency can be achieved through tools such as reading and writing quorum, so you can often achieve it in practice, even if the database is not technically consistent.

Of course, there are examples of non-CF repositories that use peer-to-peer replication, such as CouchBase , a document repository, and Riak , the KV store. These databases work very well and use automatic sorting in one form or another.

+2
source

Source: https://habr.com/ru/post/1439156/


All Articles