The downsides of using a Public Cloud provider can be summarized as follows:
- Although infrastructure in the Public Cloud (IaaS) can be consumed as would any traditional virtual machine, if a customer wants to take advantage of any other cloud service, applications may need to be rebuilt (or “refactored”) and data converted.
- The terminology, methods, and processes are different from the traditional IT environment most users would be familiar with, creating a learning curve and skills gaps. Although there may be a reduction of staff resources associated with eliminating an on-premises Data Center, new replacement resources will probably be needed to manage the resources in the Public Cloud.
- Because the financial model for the Public Cloud is operational in nature, customers need to anticipate variable and unpredictable costs for data egress, renewals, and support services. This can be mitigated somewhat by various payment methods offered by the Public Cloud providers, but typically spikes in usage will translate into an associated increase in cost. The actual cost that a customer will pay is difficult to predict in advance, and a workload may already be in production before the final operating cost is known.
- Moving the procurement of resources directly to end-user teams and away from the IT department can cause challenges in budget control and oversight.
- The opaque operational model of the Public Cloud provider model sometimes makes optimization difficult. Many aspects of performance, network, and security management historically used by system administrators to optimize their workloads are generally not available for users of the Public Cloud.
- Traditional advantages realized by techniques such as deduplication and compression of storage or overprovisioning for compute are passed to the cloud provider not to the cloud consumer.
- Typically, the resources consumed in the Public Cloud are part of a shared, multitenant hardware solution. Although the Public Cloud providers attempt to manage the complexity of this shared infrastructure, obtaining guarantees on performance and the impact of other user’s workloads require a detailed review for the specific service and associated SLAs.
- Shared resources potentially not only impact performance, but also could affect the policies and governance related to an organization’s data security and integrity.
- Public Cloud providers make available security, backup, recovery, and vaulting capabilities, but these may be expensive. In addition, these tools do not generally provide abilities to manage data at other Public Cloud provider sites or located on-premises.
- Although the Public Cloud providers describe the inherent security of their services, the responsibility and protection of the data remains with the consumer. The Public Cloud providers will refer customers to their Shared Responsibility Model; however, this varies depending on the type of services and can lead to complexity and misunderstanding.
- Because by nature the Public Cloud is accessible over the Internet, response time depends on the performance and availability of a reliable network connection between the user and the Public Cloud infrastructure. In addition, if applications are in any way latency-dependent (such as high-performance computing or Machine Learning), then the physical distance from the user to the Public Cloud provider may become a performance bottleneck.
- Although users can consume their resources as infrastructure (IaaS), in general the Public Cloud providers want users to consume their Platform services (PaaS), which is proprietary in nature and makes it difficult for users to move their workloads to another platform. This lack of portability impacts moving workloads not just between different Public Cloud providers, but also between the Public Cloud and an on-premises environment.
- The method by which users interact with the administrative functions of the Public Cloud providers’ services is typically using some sort of management console, which requires users to accumulate expertise in the details of the console. Because each Public Cloud provider has a different console, working with multiple management consoles adds operational complexity for users.
- The greatest potential downside of using a Public Cloud service is its high cost. Although the entry cost of a Public Cloud service is low and can offer an excellent return on investment, full-scale production deployments in the Public Cloud can be much more expensive than running a similar workload in an on-premises environment. In addition, due to the nature of how Public Cloud services costs are structured, users must deal with frequently unpredictable operational fees, most notably egress and ingress fees associated with data movement.