REPLICATION POLICY INSIDE VERTICA CLUSTER

Ty

Hello,

I have a cluster of 5 VERTICA nodes with the UNSEGMENTED ALL NODES.

 

1) Does that means that I have 5 replicas DB,  one on each node ?

 

2) If so,  what is the correlation with the value of K-Safety (that represents the number of replicas of the data in the database that exist in the database cluster) since it has 0, 1, or 2 as possible values with a cluster of > 3 nodes ?

Thanks for your help

Regards

 

Ty

Adrian_Oprea_1

1) Does that means that I have 5 replicas DB,  one on each node ?

 

No - it depends on the K-safety you have

 

k-safety = 0 (you don`t have copies of the data)

k-safety = 1 (you can loose 1 node and db is still up)(need a min of 3 nodes)

k-safety = 2 (you can loose 2 node and db is still up)(need a min of 5 nodes)

 

Segmented vs Unsegmented - this are projection design defintion (not realy cluster design).

 

Segmented -- large tables (spread aroud the cluster in even parts based on your segmentation key (as unique as possible - to avoid daa skew)).

Unsegmented -- small table (that can be replicated).