ARROW ENTERPRISE COMPUTING SOLUTIONS
Arrow Fusion Professional Services – VMware, vSphere
Health Check and Knowledge Transfer
Industry: Online Publishing
SHI International called on Arrow ECS Professional Services to help them bring a client’s environment into full compliance
with VMware best practices by performing a VMware vSphere Health Check Assessment and Knowledge Transfer
A long-standing Arrow ECS Solution Provider’s client had previously installed and configured hardware and VMware
vSphere software in three distinct environments: (development, integration and production) and needed to undertake an
assessment to identify possible configuration and environment issues and to mitigate the risk to the organization.
The project objective was to analyze the customer’s three VMware vSphere environments, to identify deviations from best
practices and to recommend how to best reconfigure and utilize the environment.
Arrow ECS Professional Services utilized its industry accepted ‘Analyze, Design, Plan, and Implement’ best practice model,
incorporating appropriate tools to deliver functionality, optimized system configurations and performance. This model is an
industry accepted best practice that allows for the optimal system configurations and usage of appropriate tools / function-
ality. In this engagement, an Arrow Fusion VMware Engineer performed a VMware vSphere health check and assessment.
This assessment included and in-depth analysis of the VMware vSphere 4.1 environment and associated networking and
storage systems. The in-scope tasks for this assessment included:
• Analysis of up to four (4) ESX host servers; up to 30 VMs and multiple Linux based applications
• Knowledge transfer, including operational and administrative training on the vSphere environment
• A Finding and Recommendations Document
A number of issues were identified in the VMware vSphere Health Check and recommendations were presented in light of the
current environment and proposed future environment, including:
• Networking: There were various networking constraints across all three environments. Several single points of failure
were found in the network topology with a lack of port density, causing certain environments to have ‘flat’ topologies with
all of the VMware networking elements on a single LAN segment. VMware best practices do not allow for single points of
failure at the virtual or physical network layer. Arrow’s recommendation was to increase port density by accessing the full
population of all available NICs to give the best performance and negate any single points of failure at the network layer.
In addition Arrow recommended redundancy in networking paths and components to eliminate all single points of failure
throughout the infrastructure, using redundant links as often as possible across all port groups.
ARROW ENTERPRISE COMPUTING SOLUTIONS
• Virtual Data Center: There was no redundant service console port group to use as a separate vmnic/uplink and an
alternate isolation response gateway address for more reliability in HA isolation detection. There were resource pools
and VMs as siblings in a hierarchy, which is contrary to VMware best practice and could cause unexpected performance
results. There were no vCenter Server roles, groups and permissions. There was also no Active Directory infrastructure
in place. Arrow’s recommendations were to avoid making resource pools and VMs siblings in a hierarchy in order to
avoid unexpected performance issues, to set up a redundant service console port group, using a separate VMNic/uplink,
and an alternate isolation response gateway address for more reliability in HA isolation detection.
• Virtual Machines: VMware tools were out of date or not installed for several of the CentOS images and not all guest OSs
were using the latest version of VMXNET that was supported. Arrow’s recommendation was to make sure that the latest
VMware Tools were installed, running, and up to date for all VMs, to use as few vCPUs as possible, to check to see if VMs
were using only one vCPU in an SMP VM, and avoid using virtual SMP if an application is single threaded and not benefit-
ting from additional CPUs.
• Storage: There was neither fibre-channel connectivity nor iSCSI connectivity and no backup platform for the environ-
ment. NFS shares were using the main traffic LAN via a routed network. This configuration presented the possibility
that LAN traffic to and from the virtual machines could contend with networked disk I/O operations. Arrow recommended
that NFS use a dedicated non-routed network with a limited number of hops for minimum latency and allocate separate
space on shared data stores for templates and media/ISOs from data stores for VMs. VMs were sharing the same data
stores as templates and media ISOs. Arrow’s recommendation was to set aside specific storage for media ISOs and to
move the templates and ISOs into the new storage area.
For more information about Arrow Fusion call 877.558.6677 or visit http://ecs.arrow.com/services/professional_services.html
—For internal use only—