Bare metal vs. virtual machine cloud: Which option is right for me?

Share this post:

There is no doubt that cloud brings us an entire new set of value propositions for enterprise computing environments, offering a huge set of benefits like application scalability, operational flexibility, improved economies of scale, reduced costs, resource efficiencies, agility improvement and more. It is harder to find all these benefits when considering just the traditional computing model.

Managed-Hosting-as-a-Service-MHaaSGenerally speaking, organizations today can choose between two types of clouds. Traditional clouds offer virtual machines (VMs) that are extremely easy to use but abstract disk, memory and CPU and come with a performance penalty. Bare metal clouds are essentially physical servers that can be deployed on demand and billed hourly.

When I first heard about bare metal in a cloud context, all of a sudden a question came up in my mind: From a conceptual perspective, does a cloud configuration require some kind of virtualization?


(Related: Three key advantages of using SoftLayer for cloud deployment)

In fact, the National Institute of Standards and Technology (NIST), whose cloud definition is widely accepted in the industry, omits virtualization as a criterion for cloud. The essential characteristics mentioned by NIST include on-demand self-service, broad network access, resource pooling, rapid elasticity and measured service, but not virtualization. This may surprise many in the IT community (like me!) who have always assumed that a virtualized server infrastructure was necessary to provide the flexibility and scalability associated with cloud.

With that said, you can see that VMs are great for some things and bare-metal servers are great for other things. Sometimes those reasons are technology and sometimes those reasons are based in finance issues.

You should consider VM cloud for highly dynamic workloads where you’re not really worried about performance. For example, if you have an application that spins up and down rapidly, or if you have very temporary workloads that must be spun up in minutes and run for a short period of time before you turn them off, virtualization is great.

But if your application is sensitive to performance, bare metal can be unbeatable. Bare metal cloud services refer to a type of high-performance, hosted-server offering that combines the advantages of traditional dedicated servers with those of infrastructure as a service (IaaS) virtual servers.

With bare metal cloud, all of the resources of physical servers are dedicated to a single user and can offer better performance than a comparable virtualized server. Because resources are not being shared, no hypervisor layer is needed, allowing more of the server’s processing power to be allocated to the application.

Some of the advantages of bare metal cloud services are:

• Resources dedicated to a single customer
• Greater processing power and input/output operations per second (IOPS)
• More consistent disk and network I/O performance
• Quality of Service (QoS) that guarantees elimination of the noisy neighbor problem in a multitenant environment.

Another benefit from bare metal is security. The potential of breaking regulatory compliance in a multitenant environment was the main reason why security-sensitive organizations were originally reluctant to move their data to the cloud. With bare-metal servers, it is possible to implement physical segregation of resources.

Another important issue addressed by bare metal is that there are some software vendors with license and support agreements that don’t support sub-capacity license models. In this case, the vendor does not allow the use of soft partitioning as a means to determine or limit the number of software licenses required for any given server.

You will have to pay for the software license as if you are using the entire physical server, even if you are not.

There are just a few cloud providers that offer a real bare-metal cloud computing instance in their cloud service portfolio. SoftLayer, an IBM company is leading the way in this type of offering by giving its clients the opportunity to enjoy the flexibility, provisioning and on-demand billing advantages of cloud computing and running applications on dedicated physical servers at same time.

In addition, SoftLayer customers can move data with the click of a button from virtual servers to physical and back, whenever they choose, in the SoftLayer data centers.

As you can see, there is not a correct answer for the question in the title of the post. It will always depend on the type of workload you need to run on top of the cloud.

I would like to hear from you. Which cloud would you choose and why?

Add Comment
One Comment

Leave a Reply

Your email address will not be published.Required fields are marked *

Nathaniel Crokcer

Wilson, as pointed out in the NIST definition of cloud (SP 800-45), there is no mention of virtualization, only encapsulation and an abstraction layer. Therefore it is impossible for me to answer your question.

I could choose Public Cloud, Private, Cloud, Hybrid Cloud or Community cloud, and in every one of them, I could have bare metal servers or virtualized servers.

If we truly follow NIST’s architecture, then the compute, storage and network resources should all be below my abstraction layer, meaning I shouldn’t have to deal with them. As a customer, I can specific what levels of service, including security and governance. My requirements would then determine what resources I need.

But since you asked, what would “I” choose? The answer is simple–baremetal. When I founded Bladeworks in 1999, there was no enterprise hypervisor except on system Z, so we built our cloud all on physical blades with a rapid provisioning system where all images were under version control for rapid roll-back.

I believe you can deliver all of the five requirements of NIST’s cloud, plus their architectural framework without SERVER virtualization. Storage virtualization is something I would use.

Lastly, I would say that the reason server virtualization initially became so popular for three reasons. 1. Windows DLL conflicts were out of control and there was no real HAL. Creating VMs solved this problem. 2, the rapid exponential profliferation of Intel machines resulted in servers running at 12% utilization or less. Combining VMs on a single hardware platform created the famous 10:1 ROI. 3, There was no good portability means for Windows like there was for UNIX, Mini and Mainframe platforms, where you could easily backup the OS, data and applications is a one fell swoop. The OVF revolutionized this for Windows making it finally possible to back up a server.

The unitended consequence was that it created portability like a JAVA VM, making provisioning almost effortless.

But all of those things can be easily solved with a proper bare metal architecture, eliminating the cost and performance overhead of virtualization.

Don’t get me wrong, I like Virtualization at large, just not server virtualization.

More Infrastructure stories

IBM and VMware partnership unlocks new value for enterprises

Several years ago, IBM and VMware set out to tackle two pressing industry challenges. The first was to help clients extend existing VMware workloads from on-premises environments to the cloud without having to retool operations, rearchitect applications and redesign security policies. This would help our enterprise clients seamlessly run their VMware workloads in the IBM […]

Continue reading

IBM brings the ease of containers to complex workloads with managed Kubernetes on bare metal

Today, IBM will become the first major cloud provider to enable developers and data science teams to create and run Kubernetes containers as a managed service directly on bare metal cloud infrastructure. This will widen the potential of Kubernetes, along with the significant agility and flexibility it brings to data, to apps and workloads that […]

Continue reading

How LSPI migrated its SAP workloads into IBM Cloud in seven months

Who would you turn to if you were asked to stand up your own IT infrastructure in just a few months? That’s a question my company, LiquidPower Specialty Products Inc. (LSPI), a Berkshire Hathaway Company, had to answer in 2017. LSPI is the global leader in the science of drag reduction. Specializing in maximizing the […]

Continue reading