Home > Blogs > Virtual Reality

VMware Bests Red Hat In OpenStack Performance, Cost Study

While the operating assumption is that the OpenStack framework works best on open source components such as KVM, a just completed study by Principled Technologies and commissioned by VMware showed otherwise. Tests showed remarkably higher performance and substantially reduced costs when using OpenStack with VMware technology including vSphere when compared to OpenStack with Red Hat components.

vmware openstack summary chart

In the study, OpenStack services were used to provision and manage the test configurations. The study equipment was identical except when published recommendations mandated a change. The test results showed:

  • VMware Virtual SAN (VSAN) provided 159% more IOPS than Red Hat Storage Server (GlusterFS)
  • A Cassandra NoSQL database installation performed 53% better on vSphere than on Red Hat KVM
  • Over 3 years, the total cost of infrastructure hardware and software was 26% lower on VMware than on Red Hat 

 The study recognized two trends in enterprise computing:

  • The emergence of hyper-converged architectures that can increase performance and lower costs associated with a virtualized infrastructure by having compute, network, and storage coexist closely on physical resources.
  • An interest in the OpenStack API framework as a way to provide efficient self-service provisioning and consumption of these underlying compute/network/storage resources to deploy applications on a large scale.

Cloud Performance

VMware innovations are helping customers get enterprise-class performance when exploring the OpenStack framework as a platform for large-scale application deployment. Among these innovations, the study showed that VMware Virtual SAN played an important role in providing performance advantages. Among the most significant findings related to VMware Virtual SAN, the study noted:

  • The use of direct-attached disks on the compute hosts brought proven benefits of shared storage in the VMware environment, such as High Availability (HA) and vMotion.
  • Tight integration with the vSphere [hypervisor]; scaled easily by adding more hosts to a cluster or more storage to existing hosts. In addition, VMware Virtual SAN can be managed directly through the familiar vCenter Server™ Web client console, alongside everything else in a VMware vSphere environment.
  • Every disk chosen for Virtual SAN storage belongs to a disk group with at least one solid-state drive that serves as a read and write cache. Additional storage or hosts added to the capacity and performance of a VMware Virtual SAN data store without disruption.

For the following tables, please refer to the full study for the complete test methodology and equipment setup.

Figure 1: The amount of YCSB (Yahoo Cloud Serving Benchmark) OPS achieved by the two solutions. Higher numbers are better.

Figure 1: The amount of YCSB (Yahoo Cloud Serving Benchmark) OPS achieved by the two solutions. Higher numbers are better.

Figure 2: The amount of IOPS achieved by the two solutions. Higher numbers are better. The workload was 70/30 R/W mix, random, and 4K block size.

Figure 2: The amount of IOPS achieved by the two solutions. Higher numbers are better. The workload was 70/30 R/W mix, random, and 4K block size.

Cost Comparison

The study showed that running OpenStack on VMware components required less hardware. Using VMware vSphere with Virtual SAN also lowered software costs. In total the study showed the 3 year costs were 26 percent lower. Because each OpenStack deployment and environment is different and support engagements vary widely from installation to installation, the costs of implementing the OpenStack framework were not included for either the VMware or the Red Hat platform.

Figure 3: Projected three-year costs for the two solutions. Lower numbers are better.

Figure 3: Projected three-year costs for the two solutions. Lower numbers are better.

Conclusion

The study concludes:

“In our testing, the VMware vSphere with Virtual SAN solution performed better than the Red Hat Storage solution in both real world and raw performance testing by providing 53 percent more database OPS and 159 percent more IOPS. In addition, the vSphere with Virtual SAN solution can occupy less datacenter space, which can result in lower costs associated with density. A three-year cost projection for the two solutions showed that VMware vSphere with Virtual SAN could save your business up to 26 percent in hardware and software costs when compared to the Red Hat Storage solution we tested.”

As an enterprise customer, you have choices when it comes to implementing an OpenStack framework. Your selections will impact the performance and overall cost of your scale out infrastructure. With this study, VMware has demonstrated significant performance gains and cost savings in an OpenStack environment.

Read the full study here.



VMware Bests Red Hat In OpenStack Performance, Cost Study

While the operating assumption is that the OpenStack framework works best on open source components such as KVM, a just completed study by Principled Technologies and commissioned by VMware showed otherwise. Tests showed remarkably higher performance and substantially reduced costs when using OpenStack with VMware technology including vSphere when compared to OpenStack with Red Hat components.

vmware openstack summary chart

In the study, OpenStack services were used to provision and manage the test configurations. The study equipment was identical except when published recommendations mandated a change. The test results showed:

  • VMware Virtual SAN (VSAN) provided 159% more IOPS than Red Hat Storage Server (GlusterFS)
  • A Cassandra NoSQL database installation performed 53% better on vSphere than on Red Hat KVM
  • Over 3 years, the total cost of infrastructure hardware and software was 26% lower on VMware than on Red Hat 

 The study recognized two trends in enterprise computing:

  • The emergence of hyper-converged architectures that can increase performance and lower costs associated with a virtualized infrastructure by having compute, network, and storage coexist closely on physical resources.
  • An interest in the OpenStack API framework as a way to provide efficient self-service provisioning and consumption of these underlying compute/network/storage resources to deploy applications on a large scale.

Cloud Performance

VMware innovations are helping customers get enterprise-class performance when exploring the OpenStack framework as a platform for large-scale application deployment. Among these innovations, the study showed that VMware Virtual SAN played an important role in providing performance advantages. Among the most significant findings related to VMware Virtual SAN, the study noted:

  • The use of direct-attached disks on the compute hosts brought proven benefits of shared storage in the VMware environment, such as High Availability (HA) and vMotion.
  • Tight integration with the vSphere [hypervisor]; scaled easily by adding more hosts to a cluster or more storage to existing hosts. In addition, VMware Virtual SAN can be managed directly through the familiar vCenter Server™ Web client console, alongside everything else in a VMware vSphere environment.
  • Every disk chosen for Virtual SAN storage belongs to a disk group with at least one solid-state drive that serves as a read and write cache. Additional storage or hosts added to the capacity and performance of a VMware Virtual SAN data store without disruption.

For the following tables, please refer to the full study for the complete test methodology and equipment setup.

Figure 1: The amount of YCSB (Yahoo Cloud Serving Benchmark) OPS achieved by the two solutions. Higher numbers are better.

Figure 1: The amount of YCSB (Yahoo Cloud Serving Benchmark) OPS achieved by the two solutions. Higher numbers are better.

Figure 2: The amount of IOPS achieved by the two solutions. Higher numbers are better. The workload was 70/30 R/W mix, random, and 4K block size.

Figure 2: The amount of IOPS achieved by the two solutions. Higher numbers are better. The workload was 70/30 R/W mix, random, and 4K block size.

Cost Comparison

The study showed that running OpenStack on VMware components required less hardware. Using VMware vSphere with Virtual SAN also lowered software costs. In total the study showed the 3 year costs were 26 percent lower. Because each OpenStack deployment and environment is different and support engagements vary widely from installation to installation, the costs of implementing the OpenStack framework were not included for either the VMware or the Red Hat platform.

Figure 3: Projected three-year costs for the two solutions. Lower numbers are better.

Figure 3: Projected three-year costs for the two solutions. Lower numbers are better.

Conclusion

The study concludes:

“In our testing, the VMware vSphere with Virtual SAN solution performed better than the Red Hat Storage solution in both real world and raw performance testing by providing 53 percent more database OPS and 159 percent more IOPS. In addition, the vSphere with Virtual SAN solution can occupy less datacenter space, which can result in lower costs associated with density. A three-year cost projection for the two solutions showed that VMware vSphere with Virtual SAN could save your business up to 26 percent in hardware and software costs when compared to the Red Hat Storage solution we tested.”

As an enterprise customer, you have choices when it comes to implementing an OpenStack framework. Your selections will impact the performance and overall cost of your scale out infrastructure. With this study, VMware has demonstrated significant performance gains and cost savings in an OpenStack environment.

Read the full study here.

2 thoughts on “VMware Bests Red Hat In OpenStack Performance, Cost Study

  1. thomas

    well.
    apart from the performance issues that filesystem based storage tech like gluster fs has with these kind of targets (most sane people would NOT use glusterfs to serve storage to a cassandra environment )

    you do manage to present a remarkable achievement, namely to circumvent the age-old “divide by zero error ”

    openstack is free. as in beer

    vmware / vcenter costs money

    please explain how vmware can be cheaper than free. vmware is handing out bags of money ?

  2. David

    Thomas, you said “please explain how vmware can be cheaper than free. vmware is handing out bags of money?”

    Perhaps you missed the part at the beginning “…commissioned by VMware” — meaning, this research is paid for by VMware.

    Apparently, the money already went to “Principled Technologies” to reach this conclusion.

Comments are closed.