Virtualization Products

HPE supports IT Business Analytics running on operating systems and databases on particular platforms, not specific hardware and software configurations. HPE supports IT Business Analytics customers who run HPE software products on supported operating systems and databases, irrespective of whether they are running transparent or virtualization solutions in their environment. HPE does not support these transparent or virtualization technologies directly. Since the providers of these technologies support a set of certified operating systems and hardware, the customer and the providers of these technologies will be responsible for any interactions or issues that arise at the hardware or operating system layer as a result of their use.

HPE will not require customers to re-create and troubleshoot every issue in a non-transparent environment; however, HPE does reserve the right to request that its customers diagnose certain issues in a native certified operating system environment without the transparent technology. HPE will only make this request when there is reason to believe that the environment is a contributing factor to the reported issue.

While IT Business Analytics is expected to function properly with these transparent technologies in place, there may be performance implications, which can invalidate HPE’s typical sizing and recommendations. Analysis must be performed within the context of the specific application to be hosted in a virtual environment to minimize potential resource overload, which can have significant impact on performance and scalability, particularly under peak load.

Planned Downtime. Planned Downtime for local system maintenance is the number one cause of downtime in physical environments. IT organizations have to schedule maintenance windows long ahead of time at inconvenient times to minimize the business impact. VMware vMotion eliminates the need to schedule application downtime due to planned server maintenance through live migration of virtual machines across servers with no disruption to users or loss of service.

Unplanned Downtime. Due to the complexity and cost of application-level clustering, only a minority of applications are typically protected with these solutions. In physical environments, that leaves the majority of applications with no protection against unplanned downtime. VMware provides a set of capabilities to provide simple and cost-effective availability for all applications running on vSphere. VMware vSphere High Availability (HA) provides cost effective, automated restart within minutes for all applications in the event of hardware or operating system failures.

For example, vMotion allows you to live migrate a VM from Host A to Host B without downtime. VMware HA provides high availability to VMs based on vMotion. For example, there are two hosts A and B in a cluster with HA enabled; If host A crashes, all VMs on that will be down too. In this case, VMware HA helps restart those VMs to Host B, so there is no service downtime.

HPE Vertica runs in the following virtualization environment:

Host:

• VMware version 5.5
• The number of virtual machines per host did not exceed the number of physical processors
• CPU frequency scaling turned off at the host level and for each virtual machine
• VMware parameters for hugepages set at version. 5.5 defaults
• IO measured by vioperf concurrently on all Vertica nodes: *
• 25 megabytes per second per core of write
• 20+20 megabytes per second per core of rewrite
• 40 megabytes per second per core of read
• 150 seeks per second of latency (SkipRead)
• Thick provisioned disk, or pass-through-storage

Network:

• Dedicated 10G NIC for each Virtual Machine
• No oversubscription at the switch layer, verified with vnetperf

Processor:

• Architecture of Sandy Bridge (HP Gen8)
• 8 virtual cores per virtual machine
• No oversubscription, dedicated full processors
• vcpuperf time of no more than 12 seconds
• 12 seconds ~= 2.2 GHz clock speed

Memory:

• Pre-allocate and reserve memory for the VM
• 4G per virtual core of the virtual machines
• When running vioperf, provide the –duration=2min option and start on all nodes concurrently.

HPE has tested the configuration above. While other virtualization configurations may have been successfully deployed by customers in development environments, performance of these configurations may vary. If you choose to run HPE Vertica on a different virtualization configuration and experience an issue, the HPE VerticaSupport team may ask you to reproduce the issue using the configuration described above, or in a bare-metal environment, to aid in troubleshooting. Depending on the details of the case, the Support team may also ask you to enter a support ticket with your virtualization vendor.