How does Data Compaction affect SnapMirror?
Applies to
- SnapMirror
- Data compaction
Answer
-
Data compaction stores more data in less space to increase storage efficiency
-
Inline data compaction acts on data chronologically, that is, as the data arrives at the storage controller. While the data is still in the controller memory, we take data chunks that would each normally consume an entire 4KB block on physical storage and we compact the chunks. With compaction, more than one chunk fits into a single 4KB physical block. Think of it as a suitcase-packing, trunk-packing, knapsack-packing type of problem. We’re able to take I/Os that are probably padded with a lot of zeros, or empty space, remove the empty space, and take advantage of it.
-
- Compacted data on the source node must be de-packed and re-packed before it is being sent to the network in preparation for a transfer to the destination. This extra workload comes with a performance penalty. However, since the efficiency is saved over the wire, there is much less to transfer compared to data with no efficiency, and therefore, the overall snapmirror transfer of the payload will complete faster.
Source |
Destination |
Result |
Data compaction enabled |
No data compaction |
Data compaction savings on source only |
No data compaction |
Data compaction enabled |
Data compaction savings on destination only |
Data compaction enabled |
Data compaction enabled |
Data compaction savings on both volumes* |
Additional Information