Tier relocation rules, restrictions, and guidelines
Rules
Performance monitoring, using both Auto and Manual execution modes, observes the pages that were allocated to DP-VOLs prior to the start of the monitoring cycle and the new pages allocated during the monitoring cycle. Pages that are not allocated during performance monitoring are not candidates for tier relocation.
Tier relocation can be performed concurrently on up to eight pools. If more than eight pools are specified, relocation of the ninth pool starts after relocation of any of the first eight pools completes.
If Auto execution mode is specified, performance monitoring can stop about one minute before to one minute after the beginning of the next monitor cycle start time.
The amount of relocation varies per cycle. In some cases, the cycle can end before all relocation can be handled. If tier relocation does not finish completely within the cycle, relocation to appropriate pages is executed in the next cycle.
Calculating the tier range values is influenced by the capacity allocated to DP-VOLs with relocation disabled and the buffer reserve percentages.
While a pool-VOL is being deleted, tier relocation is not performed. After the pool-VOL deletion completes, tier relocation starts.
Frequency distribution is unavailable when there is no data provided by performance monitoring.
While the frequency distribution graph is being created or the tier range values are being calculated, the frequency distribution graph is not available. The time required for determining the tier range values varies depending on the number of DP-VOLs and total capacity. The maximum time is about 20 minutes.
To balance the usage levels of all parity groups, rebalancing can be performed after several tier relocation operations. If rebalancing is in progress, the next cycle of tier relocation might be delayed.
For details about rebalancing, see Rebalancing the usage level among parity groups.