Help

Guidelines
● Local device rebuild rates are dictated by two factors: The available read performance of the source device, and the write
performance of the target device.
● Distributed device rebuild rates are dictated by three factors: The available inter-cluster bandwidth, the read performance of
the source device, and the write performance of the target device.
● Be aware of the potential performance impact of rebuilds on host I/O traffic. Tune the transfer-size of the rebuild
accordingly. The default transfer-size is 128KB. Choose this value carefully depending upon your requirements:
● A smaller transfer size has less of an impact on host I/O, but results in longer rebuild times.
● A larger transfer size allows rebuilds to complete faster, however they cause more of an impact to host I/O and have the
potential to starve out host I/O to a point of data unavailability if the host application is particularly sensitive.
● Dell EMC recommends that you set this to 128KB as a balance between minimal host impact and acceptable rebuild/
migration performance.
Corrective actions
● Tune the RAID- local device's or distributed device's transfer-size appropriately for your environment and requirements.
● For distributed device rebuilds, verify the WAN inter-cluster health and available bandwidth.
● Verify the source and target storage array health and performance.
● The system can process up to 25 concurrent rebuilds. If this is too many and overwhelms one or more storage arrays or
impacts host applications, scale back the number of concurrent rebuilds.
Filtering the Rebuild Status dashboard
You can filter the rebuild status display for:
● An individual cluster
● Distributed devices
● All rebuilds
This dashboard shows the following information about rebuilds. To see additional rebuild properties, click the rebuild name link to
open the properties dialog box.
● Locality — Where the rebuild is taking place. The rebuild can be distributed between the storage on two clusters or it can
take place at one of the clusters in a metro node configuration.
● Device — The device taking part in the rebuild.
● Rebuild Type — The type of rebuild. A rebuild can be a full rebuild or a logging rebuild. - A full rebuild copies the entire
contents of the source to the target. - An incremental build uses a checksum differencing algorithm to transfer only those
(chunks of) blocks that are different. - A comparison is used for thin devices to preserve their thinness. Comparison rebuilds
do an additional read request to the target device to determine whether a write is necessary (read from source, read from
target, compare data, only write if not equal). - A resync rewrites blocks that may have been affected by a director or
inter-cluster link failure, guaranteeing that the mirror legs are identical. Applies only to distributed-devices.
● Rebuilt — Shows the amount of data that has been rebuilt.
● Total — Shows the total amount of data that will be rebuilt when the operation completes.
● % Done — Calculates the percentage of this device that is rebuilt/migrated.
● Throughput Shows the rate of rebuild/,migration that this operation is experiencing based on read and write rates.
● ETA — The estimated time remaining for this rebuild based on the amount of data that has been rebuilt and the throughput
the rebuild operation is experiencing.
Finding a rebuild operation
Click
at the top of the screen to search for a specific rebuild job.
Sorting rebuild jobs
You can sort in the following ways:
● Click each column header to sort the data in the column, then click a second time to change the direction of the sort.
48
Monitoring the system