Netgear RD5D (ReadyDATA Disk Packs) – ReadyDATA Disk Packs (SATA/NL-SAS/SAS/SSD) Quick Setup Guide

Page of 43
Optimize Performance on a ReadyDATA System
12
Volume Performance and Configuration on ReadyDATA Platforms 
Compression and Deduplication on LUNs
NETGEAR recommends that you do not enable deduplication and compression on LUNs. 
This recommendation does not apply to shares.
Deduplication can come with a high performance overhead. NETGEAR recommends that 
you employ only one deduplication technology for a data set. For example, use the 
deduplication feature of your backup software application instead of the ReadyDATA’s 
deduplication, or the other way around. 
If the deduplication ratio is below 1.5, NETGEAR recommends that you disable deduplication 
to avoid performance overhead. In general, NETGEAR recommends that you start with 
compression, rather than deduplication. Compression has minimal performance overhead 
and often saves as much capacity as deduplication. If you intend to use deduplication on a 
large data set (for example, over 4 TB), NETGEAR recommends that you purchase a read 
cache SSD to avoid high performance overhead.
Deduplication
NETGEAR strongly advises against enabling deduplication on any LUN. Deduplication 
prevents storage of redundant data on a LUN. Unique data is stored only once on the LUN, 
and other instances of the identical data are removed and replaced by a pointer to the unique 
data. However, deduplication rarely increases the speed of data transfers, and it does not 
effectively save storage space for a LUN.
Whether you use a thin or a thick LUN, a LUN size is statically defined and therefore does not 
grow when the volume storage space increases: deduplication on a LUN might decrease the 
performance without delivering more storage capacity for the LUN.
For more information about deduplication on LUNs, visit 
.
Compression
Compression uses the CPU and memory of the ReadyDATA platform to compress and 
decompress data. Although compression can increase the speed of data transfers, it does 
not effectively save storage space for a LUN.
Whether you use a thin or a thick LUN, a LUN size is statically defined and therefore does not 
grow when the volume storage space increases: compression on a LUN might decrease the 
performance without delivering more storage capacity for the LUN.
The following graphs show that enabling compression on a LUN does not provide any real 
benefits. Write operations are more affected than read operations.
Drag and Drop Performance Overhead with Compression
The following figure shows that enabling compression weakens the drag and drop 
performance for both read and write operations. The two bars on the left show write 
operations. The two bars on the right show read operations.