Scaling

AIStor provides tools for scaling your cluster capacity by adding or removing server pools. These tools allow you to expand storage capacity or retire aging hardware without disrupting ongoing operations.

MinIO recommends initially sizing clusters for at least 2-3 years of anticipated data growth. This approach minimizes the frequency of expansion operations and simplifies long-term capacity management.

When planning capacity, target no more than 70% storage utilization to maintain optimal performance and allow room for data growth.

AIStor scales horizontally by adding new server pools to an existing cluster. Each server pool is an independent set of nodes, and all pools together form a single namespace for storing objects.

Key considerations for expansion:

• New pools must meet minimum erasure code requirements compatible with existing pools.
• All nodes in the new pool should have bidirectional network connectivity with existing nodes.
• MinIO recommends similar hardware configurations across pools for predictable performance.

For detailed procedures, see Expand Available Storage.

Decommissioning removes server pools from a cluster with two or more pools. This process is designed for retiring aging hardware or consolidating capacity as part of infrastructure refresh cycles.

When you decommission a pool, AIStor marks it as read-only and automatically migrates all objects to remaining active pools while continuing to serve read requests.

Key considerations for decommissioning:

• The cluster must have sufficient free capacity across remaining pools to absorb the decommissioned pool’s data.
• Complete any planned expansions before starting decommissioning to ensure balanced data distribution.
• Back up cluster settings before starting the decommission process.

For detailed procedures, see Decommission Aged Hardware.

Server pool expansion and decommissioning provide operational flexibility, but they do not provide disaster recovery protection. For production clusters requiring business continuity and disaster recovery, use site replication to maintain synchronized copies of your data across geographically distributed AIStor clusters.