Cisco Cisco UCS C460 M2 High-Performance Rack Server 白皮書
37 © 2016 Cisco and/or its affiliates. All rights reserved.
Recommended StorMagic SvSAN Mirror Configurations
Overview
To allow the VSA storage to dynamically scale with odd numbers of Cisco UCS blades, you should configure each VSA to divide
its pooled storage into two iSCSI mirrors. This configuration also adheres to VMware’s default storage heartbeat policy, which
requires a minimum of two data stores when deploying a two-blade configuration.
its pooled storage into two iSCSI mirrors. This configuration also adheres to VMware’s default storage heartbeat policy, which
requires a minimum of two data stores when deploying a two-blade configuration.
Mirrored Targets
A mirrored target is a target that is mirrored between a pair of VSAs, with each VSA hosting one side of the mirror (called a
mirror plex). Therefore, there are two identical copies of the target data when the system is running normally: one on each
VSA. Any data sent to one plex is automatically copied to the other. This copy operation is performed synchronously, meaning
that if an initiator sends data to one plex, the VSA does not acknowledge the request until the data is present on both plexes. A
synchronous mirror can be used to provide highly available storage. An initiator can access any plex at any time, and if one plex
fails. The initiator can continue without interruption by using the other plex.
mirror plex). Therefore, there are two identical copies of the target data when the system is running normally: one on each
VSA. Any data sent to one plex is automatically copied to the other. This copy operation is performed synchronously, meaning
that if an initiator sends data to one plex, the VSA does not acknowledge the request until the data is present on both plexes. A
synchronous mirror can be used to provide highly available storage. An initiator can access any plex at any time, and if one plex
fails. The initiator can continue without interruption by using the other plex.
As previously mentioned, the pool of storage on each VSA should be divided into two iSCSI mirrors (Figure 50). This approach
allows dynamic storage migration when you add blades and VSAs to the cluster. This configuration enhances the reliability of the
solution in the event of multiple blade failures, and reduces the possibility of any storage contention, by spreading multiple virtual
machines across mirrored targets.
allows dynamic storage migration when you add blades and VSAs to the cluster. This configuration enhances the reliability of the
solution in the event of multiple blade failures, and reduces the possibility of any storage contention, by spreading multiple virtual
machines across mirrored targets.
Figure 50. Four-Blade Sample Mirror Configuration
VD 2
2TB
ESXi Boot
Drive
ESXi Boot
Drive
VMware vSphere Cluster
2 x 1.2TB RAIDO
VD 1
185GB
ESXi
ESXi
VD 1
185GB
2 x 1.2TB RAIDO
VD 2
2TB
VSA
Local VMFS
VSA
Local VMFS
Shared
VMFS
Shared
VMFS
VD 2
2TB
ESXi Boot
Drive
ESXi Boot
Drive
2 x Synchronous Mirrors
2 x 32GB SD card
Hypervisor (HV)
Partition Mirror
2 x 1.2TB RAIDO
VD 1
185GB
ESXi
ESXi
VD 1
185GB
2 x 1.2TB RAIDO
VD 2
2TB
VSA
Local VMFS
VSA
Local VMFS
Shared
VMFS
Shared
VMFS