- Scott Sinclair, Enterprise Strategy Group
Multi-cloud IT environments are the new normal. According to research from my company, Enterprise Strategy Group, 81% of IaaS users have more than one cloud service provider. As a result, businesses have more choices when deploying new workloads. The public cloud is a compelling infrastructure option that offers multiple alternatives from which to choose.
With so many public cloud choices, IT organizations are developing new strategies on how and where to deploy new workloads. With public cloud adoption increasing, one particular workload deployment has also seen an increase: a cloud-first strategy.
The basic premise of a cloud-first approach is that all new workloads are deployed on public cloud infrastructure unless there is a compelling reason to keep the workload on premises. Frequently, these strategies are driven from a company's executive-level and have some sound rationale behind them.
Firms often approach the public cloud from a state where all existing applications are deployed on premises in the data center. In this case, a cloud-first strategy for new workload deployments can help balance workloads across a hybrid cloud environment by building out the public cloud side with new workloads.
A cloud-first model can help address perceptions that IT decision-makers are biased toward on premises. This is often not the case, as many have embraced the public cloud, but the perception that IT can be resistant to change -- and to the public cloud -- persists. Additionally, implementing a cloud-first strategy can speed up initial application deployment as the organization can often bypass workload-sizing activities and forgo architecting, procuring and deploying new on-premises infrastructure.
Going cloud-first called into question
While the logic behind cloud-first workload deployments has some sound elements, evidence has emerged indicating a cloud-first strategy may add unnecessary costs. In an Enterprise Strategy Group (ESG) study of IT organizations, 41% of IT decision-makers identified that they had moved a workload back from the public cloud to be run on premises. In response, ESG conducted a separate, detailed study of respondents that pulled workloads back from the cloud to understand what happened and why. The study revealed several insights, one of which calls into question cloud-first application deployment strategies.
Those that identified themselves as having a cloud-first strategy for new application deployment were three times more likely to identify that they moved many workloads back from the cloud compared to organizations that equally considered on premises and public cloud prior to deployment.
The likely explanation for this discrepancy includes these three assertions:
- There are factors that make certain workloads better suited for an on-premises or a public cloud deployment.
- These factors are not obvious and are influenced by workload, data, performance characteristics and expected growth, as well as specifics regarding the business and the IT department making the decision.
- When a firm has a cloud-first deployment strategy, not enough upfront work is done to investigate these factors. This not only increases the likelihood that the public cloud will be identified as a less-than-ideal fit, but it also increases the chances the workload will come back to an on-premises environment.
As the number of workloads pulled back from the cloud increases, cost to the business increases. Some impacts include the cost of new on-premises infrastructure, the export costs of migrating data back on premises, and the time and personnel resources involved. In addition, according to ESG research, 43% of IT organizations that moved a workload back from the cloud reported incurring costs related to downtime while the application was migrated back.
These costs are only part of the overall impact. In this same study, 84% of IT decision-makers that pulled a workload back from the public cloud also indicated that they were either much less likely or somewhat less likely to use public cloud services in the future. In other words, each workload an organization pulls back sours IT on the public cloud experience, making it less likely to use the public cloud in the future. That future workload -- the one that doesn't get deployed in the cloud -- might have been one that would benefit from a public cloud infrastructure model.
Going hybrid often the optimal solution
Some observers might argue that since the study included IT decision-makers, the choice to take a cloud-first strategy might have come from the executive leadership and would therefore stay in place. This could, and likely does, occur in a percentage of these environments. However, friction would still be created among IT, line of business and executive teams, likely hindering future productivity.
Either way, the response to all these costs is not to take an on-premises-first approach. Public cloud solutions offer too many benefits to be dismissed entirely.
The likely optimal solution is to adopt a hybrid-first model, one that equally considers both the public cloud and on premises prior to deployment. A crucial attribute to making this model a success is defining the decision criteria upfront and building a repeatable model that can be executed quickly and with minimal bias. While models will likely vary based on the company, elements that should be included in any evaluation are data sensitivity and regulatory compliance requirements, in addition to the workload's performance and growth characteristics.
A successful hybrid cloud-based business does not result from having two separate factions -- one supporting on premises and the other supporting public cloud -- battling it out, or even from top-down executive mandates. Success results from collaboration and building a systematic, repeatable model for workload deployments, one that expects some bias but focuses on objective metrics when making a hybrid cloud decision.