An S Series Node that has five or more available enclosures with data drives stores new objects with an optimal distribution of data and parity chunks. Optimal distribution means that no enclosure has more than six chunks for the data associated with any given object.
With fewer than five enclosures with data drives in the S Series Node, the distribution of the data and parity chunks for new objects is not optimal. Without optimal distribution, the S Series Node cannot guarantee object availability if an enclosure becomes unavailable.
When you add a fifth enclosure to an S31 Node or a sixth enclosure to an S32 Node, the S Series Node automatically optimizes the distribution of data and parity chunks. The optimization process is similar to the repair process. Object chunks are removed from the enclosures with too many chunks and re-created on the new enclosure.
Object repair has priority over the optimization process. If a repair backlog exists, optimization does not occur until the backlog is gone. However, because object repair entails the relocation of object chunks, repair can result in the optimal distribution of those chunks.
Optimization is a resource-intensive process and can significantly slow data ingest. If this degradation becomes an issue, your authorized service provider can make changes to the optimization process to lessen the impact or temporarily stop the process.