Public cloud services have become so simple to purchase and deploy that many organizations are failing to do basic due diligence on how much these services really cost and whether they can deliver the performance required for key workloads. This can be a big mistake, according to a report by Frost & Sullivan’s Stratecast practice, negatively affecting budgets and business operations.¹ Many IT leaders are discovering flaws in their initial cloud purchase criteria, leading to:
- Higher than expected costs
- Unacceptable application performance
- An inability to discern whether they are overpaying for the services they are receiving
IT teams can avoid these problems by better understanding what they are—and are not—purchasing from public cloud providers. As described in the Frost & Sullivan report, here are six of the most common and potentially harmful myths that could derail your organization’s cloud initiative.
Myth No. 1: If it’s not virtualized, it’s not cloud
There’s a widespread belief that cloud benefits and economies are possible only in a virtualized infrastructure. This is a misconception that can keep you from exploring bare metal or physical server options that may enable certain workloads to work better—at a lower cost—than virtualized options. Bare metal servers provide the highest levels of raw throughput for high-performance workloads, as well as flexibility to configure storage and network resources. As such, you should include a bare metal option in your price-performance assessment for many workloads.
Myth No. 2: Cloud is always cheap
If you match your workload to the right cloud service, you will probably achieve price-performance levels beyond what you could achieve with an on-premises solution. But be aware that the right cloud service for each workload may not be a public or virtualized cloud. Depending on the workload, the optimal solution may involve bare metal or dedicated servers, or a private or hybrid cloud model. If you don’t take the time to ensure alignment between the cloud deployment model and service, you could easily spend much more money than necessary.
Myth No. 3: The lowest per-unit rate for compute and storage provides the best overall value
Each cloud service provider designs and prices its cloud services uniquely. Some infrastructure and service components may be prepackaged into bundles of capacity, while others may be available à la carte. To determine what you will pay to run your workload, you must understand how each provider prices each component. One example: Some providers charge a per-GB fee for intra-cloud data transfer, while others include unlimited transfer at no charge. In each case, the actual charges to run a workload may be many times higher than the basic per-unit rates might indicate at first glance, particularly for network-dependent workloads.
Myth No. 4: All CPUs are alike
Cloud service providers enable customers to select the right-size compute unit based in part on processor capacity, generally referred to in “cores” or virtual CPUs. In evaluating price-performance, remember these key points:
- A virtual CPU is not the same as a physical CPU. In a virtualized environment, your workload does not have access to the full core listed in the description because some processor capacity will support the hypervisor.
- You have to compare equivalencies. Many cloud service providers use Intel’s Hyper-Threading Technology to maximize processor throughput. With this technology, a single physical core can support two hyperthreads, allowing the provider to sell two virtual CPUs for each physical core. In comparing bare metal cloud servers to virtualized servers of the same processor generation, assume two virtual CPUs are roughly equivalent to one physical core.
- All virtual cores are not alike. Within a virtualized cloud environment workload performance will be impacted by the specific processor hardware on which it is running. A cloud workload running on the latest generation Intel Haswell processor delivers much higher performance than the same workload running on older generation hardware, such as Sandy Bridge. It’s worth investigating the actual or likely processor as you research your cloud service options. Some cloud providers charge higher rates for newer generation hardware.
Myth No. 5: Cloud is simple and fully self-serve
Most businesses don’t have the staff expertise, time or available resources to successfully deploy and maintain a cloud environment. Consequently, many IT decision makers face unexpected challenges and expenses after beginning their cloud implementations. If you underestimate what it will take to achieve the benefits promised by cloud, you expose yourself and your organization to unnecessary risk. According to the Frost & Sullivan Stratecast report, 31% of IT decision makers surveyed said they have struggled to attain additional budget to fully implement their cloud strategies. In addition, 21% said they were challenged to gain senior leadership support to continue executing on those strategies.
Myth No. 6: Performance is fairly similar across cloud providers’ services, so price should be the definitive selection criterion
Cloud is not a commodity. If you deploy the exact same workload into different providers’ equivalent cloud services, performance can vary significantly—thanks to differences in the way providers configure their clouds, the infrastructure components they choose, the procedures and protocols they follow, and other factors. By considering price alone, rather than price for equivalent performance, you risk selecting cloud services that are not optimal to run your workloads. Furthermore, the problem can spiral, since the tendency in dealing with lower than acceptable workload performance is to spin up additional cloud servers—unnecessarily driving up costs.
Conclusion
Cloud computing is changing the nature of how IT is purchased, delivered and consumed. Public cloud services provide huge opportunities for organizations to save money and increase business agility. But for IT teams, it is critical to understand how to best leverage the benefits of the cloud so they can deliver the right performance to the right workloads at the right price. Understanding the common myths about cloud price-performance is an important step in that direction. As the saying goes, a little knowledge goes a long way.
¹“The Truth about Cloud Price-Performance: How Misperceptions about Service Costs Can Derail Your Cloud Strategy,” Frost & Sullivan Stratecast, August 2015
