You can reduce redundancy on a cluster by running SmartDedupe. Deduplication creates links that can impact the speed at which you can read from and write to files. In particular, sequentially reading chunks smaller than 512 KB of a deduplicated file can be significantly slower than reading the same small, sequential chunks of a non-deduplicated file. This performance degradation applies only if you are reading non-cached data. For cached data, the performance for deduplicated files is potentially better than non-deduplicated files. If you stream chunks larger than 512 KB, deduplication does not significantly impact the read performance of the file. If you intend on streaming 8 KB or less of each file at a time, and you do not plan on concurrently streaming the files, it is recommended that you do not deduplicate the files.
Deduplication is most effective when applied to static or archived files and directories. The less files are modified, the less negative effect deduplication has on the cluster. For example, virtual machines often contain several copies of identical files that are rarely modified. Deduplicating a large number of virtual machines can greatly reduce consumed storage space.