How to Choose an Enterprise Grade Public Cloud

With VMware having announced the VMware Cloud on AWS Bare Metal, a new segment of the public cloud business appears to be emerging which is focused upon meeting the needs of enterprises with applications that need to be up all of the time, and that need to perform well all of the time. In a previous post we speculated how the industry might transform itself to address this new opportunity. In this post we suggest some criteria that enterprises should use to chose an enterprise grade public cloud.

Choosing your  Enterprise Grade Public Cloud

There will likely be many combinations of bare metal public clouds (we know that Amazon, IBM, and Rackspace offer this option, and there will likely be many others). Here is a suggested set of criteria that you should use to help make your choice:

  • Seamless integration with your existing private and hybrid cloud – If you are running VMware vSphere for your own private or hybrid cloud, then you should be able to set up the networking so that the bare metal public cloud simply appears as an extension of your existing environment. In fact you should be able to vMotion VM’s back and forth between your on premise cloud and your bare metal public cloud.
  • Single Tenancy – The point of a bare metal cloud is so that you can guarantee that your workloads and applications get the resources that they need to deliver excellent performance and to ensure that you should be able to get dedicated servers with no one else’s workloads running on the same servers as yours. This might go so far as also requiring dedicated networks and storage devices.
  • Co-location of Bare Metal Workloads – Today, many enterprises run their database servers on bare metal, and run everything else virtualized in vSphere. If the bare metal cloud provider offers the ability to run certain workloads natively on the hardware next to the Cloud OS, then this would allow customers to keep their existing operating model for these performance and availability critical workloads.
  • Elasticity – It would be really nice if the single biggest benefit of the shared tenant public cloud, the ability to scale up and down – was also present in the single tenant bare metal cloud. This would mean that if you need additional servers in your bare metal cloud then it would be a matter of software provisioning to get this additional capacity upon demand.
  • Elastic Pricing – It would also be very nice if the bare metal cloud had consumption based pricing so if you were not using your servers at night you did not have to pay for them. This should also apply to bursting for additional capacity. Ideally, the bare metal cloud should burst in terms of capacity and pricing just like a shared tenant public cloud.
  • Full Support of Cloud OS Features – this is a tricky one, because it is going to depend upon which Cloud OS you choose. If you pick VMware this should mean that all of the valuable features that come with vSphere come along for the ride including – vMotion, DRS, High Availability, Fault Tolerance, Site Recovery Manager, etc.
  • Metric Visibility through the Entire Infrastructure – This is what will determine how much these new bare metal clouds can eat into the existing on premise estate of data center hardware. If you own your own hardware you get the metrics from the storage arrays, the network devices and the physical servers. If these new bare metal clouds are really going to support the kinds of workloads that run on premise today, the bare metal cloud provider is going to have to be fully transparent as to the operation of the entire hardware infrastructure stack.
  • Support for your existing management tool set. If you have chosen backup tools, security tools, capacity tools, performance tools, monitoring tools, whatever – if those tools work in your on premise private cloud they should work if you pick the same Cloud OS in your bare metal cloud. Let’s start with vCenter. You should be able to use your vCenter to seamlessly manage hosts, guests and data stores across internal and public clouds. All of the products that use various VMware API’s should simply and transparently work.
The OpsDataStore Strategy

At OpsDataStore we combine metrics from VMware vSphere, AppDynamics, Dynatrace, ExtraHop and Intel into end-to-end topologies which allow for dramatic improvements in end-to-end troubleshooting and real time capacity management. We are hopeful that the above requirements will be met by multiple combinations of bare metal cloud providers and Cloud OS providers so that customers can use our ecosystem of vendors to manage their private, their hybrid, and their public clouds.


The VMware Cloud on AWS looks very promising. We know from our work with large enterprises what demands they place on their internal clouds. They have these on-premise clouds for good reasons. If the bare metal clouds are going to make an impact they must offer cloud benefits (elastic operation and pricing) while not taking away any of the enterprise grade features that customers rely upon today.

For More Information Visit These Links: