node-replication
node-replication copied to clipboard
An operation-log based approach for data replication.
This addresses [issue 52](https://github.com/vmware/node-replication/issues/52). We've added `add_replica` and `remove_replica` to the `NodeReplicated`. Thread context is held within the replica . The number of new replicas are limited via `MAX_REPLICAS_PER_LOG`.
### Is your feature request related to a problem? Please describe. Static replication can consume too much memory. Currently CNR/NR set-up the replication factor at data-structure instantiation and doesn't allow...
### Is your feature request related to a problem? Please describe. Partitioning a DS leads to more write-scaling however the write-scaling is bottlenecked by the amount of replication we do...
The [async](https://github.com/vmware/node-replication/pull/21) variant for NR works but the read throughput is still much lower than non-async (30% drop in perf). Likely this is due to necessary allocations.
I think CNR is also affected by the same bug, we should apply it in both libraries.