Scenario:
2 server rooms 1km apart with direct fiber connection.
One 10 node/rack in each server room.
What would make most sense?
1. Create two clusters with sync rep.
2. Create one cluster with rack fault tolerance (one rack in each server room)
/Peter
Not sure what your actual bussiness requirments are for up time but solution 1 is better overall. 2. Also wouldn’t product from a site failure if 2/3 nodes when down on a side. There are some techincal design issues with rack awarness across sites due to how we handle the metadata.
Scenario:
2 server rooms 1km apart with direct fiber connection.
One 10 node/rack in each server room.
What would make most sense?
1. Create two clusters with sync rep.
2. Create one cluster with rack fault tolerance (one rack in each server room)
/Peter
Hello
1-Rack fault domain needs minimum 3 rack not 2 for RF2.
2-Sync Replication makes more sense depend on your RPO objectives
3-Check the below link for Sync Replication requirements:
Enter your E-mail address. We'll send you an e-mail with instructions to reset your password.
