Cluster data usage grows and occasionally it grows more rapidly then planned and expected before a new node to be added to handle this growth.
Nutanix offers data Compression and Deduplication to hold more data in the container, by reducing the stored size and avoiding duplicate data, respectively.
Usually Compression should be used first, as de-duplication is recommended only on some specific scenarios (please, check documentation links provided below).
You can enable compression on a storage container. Compression can save physical storage space and improve I/O bandwidth and memory usage which may have a positive impact on overall system performance.
The following types of compression are available.
- Post-process compression: Data is compressed after it is written. The delay time between write and compression is configurable, and Nutanix recommends a delay of 60 minutes. If compression is enabled in "Post-process", then existing data will also be compressed after some time (when curator scans are completed).
- Inline compression: Data is compressed as it is written and only new data will be compressed
For some more info about compression please see the link below:
Deduplication reduces space usage by consolidating duplicate data blocks on Nutanix storage. You can enable either cache deduplication only or both cache and capacity deduplication on a storage container.
You can enable cache deduplication of read cache to optimize performance. Cache deduplication means deduplication performed on the data in memory and in solid state storage (SSD).
You can enable capacity deduplication of persistent data to reduce storage usage. Capacity deduplication means deduplication performed on the data in hard disk storage (HDD).
Note that you can enable capacity deduplication only if cache deduplication is enabled.
Vdi full clones and persistent desktops can make best use of this feature.
For some more info about compression please see the link below
The cost consideration:
The amount of metadata to manage these features is higher and the growth of metadata is not balanced across the nodes in the cluster, the concentration of growth on individual nodes can trigger compression of fingerprinting to be disabled on the cluster causing the cluster space usage to fill up the container; so attention should be given to balancing this when enabling these features. This is specially visible in storage heavy nodes (non-compute nodes). To prevent this the internal parameters of the AOS should be set to appropriate values for different nodes to prevent this from happening. Of course this can only be done by requesting assistance from Nutanix support.
For a more complete review of Storage components and features that includes “Erasure Coding” feature please review: