Information Technology Reference
In-Depth Information
Figure 6.1
When ESXi hosts
are connected to
that same shared
storage, they share
its capabilities.
Shared Storage
When sizing or designing the storage solution, you focus on attributes like capacity (giga-
bytes or terabytes) and performance, which is measured in bandwidth (megabytes per second,
or MBps), throughput (I/O operations per second, or IOPS), and latency (in milliseconds). It goes
sometimes without saying, but designing for availability, redundancy, and fault tolerance is also
of paramount importance.
Determining Performance Requirements
How do you determine the storage performance requirements of an application that will be virtual-
ized, a single ESXi host, or even a complete vSphere environment? h ere are many rules of thumb
for key applications, and the best practices for every application could fi ll a book. Here are some
quick considerations:
Online transaction processing (OLTP) databases need low latency (as low as you can get, but a
few milliseconds is a good target). h ey are also sensitive to input/output operations per second
(IOPS), because their I/O size is small (4 KB to 8 KB). TPC-C and TPC-E benchmarks generate
this kind of I/O pattern.
◆
Decision support system/business intelligence databases and SQL Server instances that support
Microsoft O ce SharePoint Ser ver need high bandw idth, which can be hundreds of megabytes
per second because their I/O size is large (64 KB to 1 MB). h ey are not particularly sensitive to
latency; TPC-H benchmarks generate the kind of I/O pattern used by these use cases.
◆
Copying fi les, deploying from templates, using Storage vMotion, and backing up VMs (within
the guest or from a proxy server via vSphere Storage APIs) without using array-based approaches
generally all need high bandwidth. In fact, the more, the better.
◆
