Is there rack awareness in nebula graph? #4256
-
|
In for example Apache Cassandra there is a rack awareness concept (using the network topology strategy). The purpose of this is to "mark" in configuration which rack/chassi each server runs on. For example, assume a 9-node cluster: Rack1 Rack2 Rack3 If we use this setup and then apply replication_factor=3 in our keyspace we can be sure that each piece of data is replicated among all three racks. This means that not only can we survive temporarily losing a server or two, we can survive losing all nodes in an entire rack/chassi. Is this possible to achieve with Nebula Graph? |
Beta Was this translation helpful? Give feedback.
Replies: 1 comment 1 reply
-
|
Good point. There used to be a feature called "zone" in Nebula 2.6 very similar to this idea. Check it out https://docs.nebula-graph.io/2.6.2/7.data-security/5.zone/ |
Beta Was this translation helpful? Give feedback.
-
|
Yes, the term zone used to be introduced(and removed T_T) was to address the different levels of fault sets like the rack, chassis, etc. |
Beta Was this translation helpful? Give feedback.
Good point. There used to be a feature called "zone" in Nebula 2.6 very similar to this idea. Check it out https://docs.nebula-graph.io/2.6.2/7.data-security/5.zone/
However, the zone feature has been removed from 3.x since it's not mature enough. It might be brought back in the future but no timeline yet.