White Papers

6 Dell Storage PS Series Arrays: Scalability and Growth in Virtual Environments | TR1072

2 Virtual, hypervisor-based environments

Virtualization adds a new set of planning and scaling challenges to the data center. For example, the

advanced virtual machine availability features in VMware, such as High Availability, Distributed Resource

Scheduling, Fault Tolerance, and VMware vSphere

Storage vMotion

, depend on the ability of all the

VMware ESXi™ hosts in a cluster to see all of the volumes. The same is true for Microsoft

Hyper-V

clusters

utilizing Clustered Shared Volumes. Many customers start their virtual environments small. However, because

these environments can scale easily by adding new servers and more storage, this often results in a redesign

to accommodate the growth of the environment. This growth and scalability can also leave customers in

situations where the growth of the virtual server environment exceeds the allowable active connections limits

to the PS Series pool or Group.

For clusters in general, the formula for calculating the connection count to a PS Series pool is:

Total connections = hosts x connections per volume (sessions) x volumes x PS Series members in pool

Note: The number of VMs does not have anything to do with the connection-count formula but overhead

must be taken in account for things like iSCSI-initiated connections from VMs, volume snapshots, or backup

volumes.

That number is compounded when MPIO technologies like the PS Series Device Specific Module (DSM) for

Windows

or the Multipathing Extension Module (MEM) for VMware vSphere

are enabled. These advanced

MPIO modules install a connection management service with intelligent path-selection capabilities that

maintains a table of contents on each host of all the relevant volumes on the PS Series Group members. This

table of contents allows the MPIO service on the host to know exactly where the data blocks for an attached

volume reside on the SAN and the path-selection policy assures the appropriate path is used to access the

data. This allows for extremely efficient least queue depth MPIO with performance benefits.

The DSM for Windows and MEM for vSphere MPIO modules are based on the same algorithm that will, by

default, create two connections per volume slice (volume slice = distribution of volume data across arrays in

the same pool) supporting up to six total connections per volume. When using the PS Series DSM and MEM,

it is important to understand how volume connections are created and managed. All PS Series MPIO modules

are built to provide performance benefits by creating connections to all members that a volume spans,

thereby providing a direct path from the host to the data being requested.

See Configuring and Deploying the Dell PS Series Multipath I/O Device Specific Module with Microsoft

Windows.

See Configuring and Installing the PS Series Multipathing Extension Module for VMware vSphere and PS

Series SANs.