Hi, Is it possible to direct write request from application to a particular node in cluster ? As per the documentation “Data replication is configurable on a bucket-level and node-basis”.
For Example in 5 node cluster, I want my application writes for a particular logic inserts data in Node 1. This will make my Node 1 to be fixed for that particular operations.
I understand, if possible, this will lead to major dependency and risk in case node 1 goes down…but I need that way.
No this is not possible. The destination node is determined at the protocol level, by a hashing of the key. For scalability and availability issues, you cannot change that (and as you say this would be dangerous).
If I’m not mistaken, the documentation extract you cited meant to say that the replication factor can be tuned at the bucket level (tune the number of nodes on which data will be replicated, not choose the nodes themselves).
Perhaps you could tell a little bit more about why you think you need it that way? I’m sure there is another solution to your problem.