Part 3 of 5 · 1 min read
Automatic versus manual rebalancing
A system can decide on its own when and how to move data, or require an operator to trigger and direct each move.
Think of a thermostat that adjusts the temperature on its own, without anyone touching a dial. Compare that to a furnace that only changes output when someone deliberately turns a knob.
In plain words: automatic rebalancing lets the database decide on its own when data needs to move and carry out that move without an operator's direct involvement, reacting to changes in cluster size or load without manual intervention. Manual rebalancing instead requires an operator to trigger the rebalancing process, and often to decide which specific data moves where, giving more control at the cost of needing a human to notice the need and act on it. Automatic rebalancing is more convenient, but it can make an unexpected or poorly timed decision, such as rebalancing during a brief, temporary spike in load that would have settled on its own.
A worked example. A cluster using automatic rebalancing notices a new node has joined and immediately begins moving data to it with no operator action. A cluster using manual rebalancing instead waits for an operator to run a rebalancing command once they've confirmed the new node is healthy and ready.
The key lines. No new SQL; whether rebalancing is automatic or manual is a cluster-level configuration choice in the database's operational tooling.
Automatic rebalancing, manual rebalancing
Automatic rebalancing lets the database trigger and carry out data movement on its own. Manual rebalancing requires an operator to trigger, and often direct, the process.
Example: a self-adjusting thermostat is automatic; a furnace that only responds to a hand-turned knob is manual.