White Papers

Design best practices
10 Dell EMC SC Series: Microsoft Windows Server Best Practices | 680-042-007
2 Design best practices
This section provides guidance on sizing and configuration options for SC Series storage and Windows
Server.
2.1 Right-size the storage array and fabric
Optimizing performance is a process of identifying and mitigating design limitations that cause bottlenecks —
the point at which performance begins to be impacted under load because a capacity threshold is reached
somewhere within the overall design. The goal is to maintain a balanced configuration across the design that
allows the workload to operate at or near peak efficiency.
Before deploying a new SC Series storage array, it is important to consider the environmental design factors
that impact storage capacity and performance so that new or expanded storage is right-sized for the
environment. If the SC Series array will be deployed to support an existing Windows Server workload, metrics
such as storage capacity and I/O demands might already be understood. If the environment is new, these
factors need to be determined to correctly size the storage array and the storage fabric.
Many common short- and long-term problems can be avoided by making sure the storage part of the solution
will provide the right capacity and performance in the present and future. Scalability is a key design
consideration. For example, Windows Server clusters can start small with two nodes, and expand one node at
a time, up to a maximum of 64 nodes per cluster with newer operating systems. Storage including SC Series
arrays can start with a small number of drives and expand capacity and I/O performance over time by adding
expansion enclosures with more drives, along with additional front-end ports on supported models, as
workload demands increase. With
SC Series Live Migrate and Live Volume, up to 10 similar or dissimilar
arrays can be clustered as part of a federation to support horizonal and vertical scaling.
One common mistake made when sizing a storage array is assuming that total disk capacity translates to disk
performance. Installing a small number of large-capacity spinning drives in an array does not automatically
translate to high performance just because there is a lot of available storage capacity. There must be enough
of the right kind of drives to support the I/O demands of a workload in addition to raw storage capacity. With
SC Series, customers can choose between spinning, hybrid, and all-flash arrays depending on the needs of
the workload.
Another common mistake is incorrectly sizing (or failing to anticipate the I/O demand on) the storage fabric. A
high-performance storage array will not perform to expectations (and may appear to be the performance-
inhibiting culprit erroneously) if the speed or capacity of the FC or SCSI fabric (HBAs, NICs, switches)
severely limits I/O throughput because the fabric design is significantly undersized for the storage array. If a
storage array or storage federation is expected to generate a certain level of IOPS, make sure the fabric is
right-sized to support this level of I/O also.
Where available, customers can confidently use the configuration guidance in Dell EMC storage reference
architecture white papers as good baselines to right-size the design of their environments. See
SC Series
technical documents and videos for a list of reference architecture white papers.
Work with your Dell EMC representative to complete a performance evaluation if there are questions about
right-sizing an SC Series storage solution or fabric for your environment and workload.