The budget items and calculations for a on-premises data center are dramatically different when compared to those of cloud computing. Many of the costs of traditional on-premises infrastructure are taken care of by the cloud provider. Your organization is essentially renting the staff, compute hardware, and other infrastructure from the provider, cutting down on capital expenses, though gaining new operational expenses, that is, your cloud bill. You are also gaining the ability to scale the data center as needed.
Most people think of server racks, network-attached-storage, routers, switches, etc., when they think of a data center. However, standing up an on-premise data center requires a large amount of support infrastructure. Picking an empty room and filling it with racks of computers will eventually lead to large headaches and downtimes (not to mention safety or code violations). This additional infrastructure for an on-premises data center includes such things as climate control, fire suppression systems, security infrastructure, and a host of other considerations. Data centers also require a powerful network connection with a fast up-speed if they are to be accessed off-site.
To minimize downtime, the electrical infrastructure of an on-premises data center needs to far exceed what might be acceptable in a standard office. Backup generators, uninterruptible power supplies, and additional grounding are other costs that cannot be ignored. In a cloud model, these large upfront infrastructure costs are handled the cloud service provider. Again, as the large capital expenses of compute, networking, and storage are also handled by the cloud provider, hosting a data center in the cloud will move the IT budget toward additional operational expenses in exchange for greatly reduced capital expenses.
There are also costs associated with an on-premises data center that don’t need to be budgeted for in a cloud setup, such as the duplication of support infrastructure, and other redundancies to avoid most catastrophic failures, which are are difficult to budget for with an on-premises data center. In the simplest example, a warm off-site duplicate of your entire infrastructure would help guarantee uptime, but is extremely cost-prohibitive. On-premises data centers also require dedicated IT staff, such costs are partially externalized to the provider when running a data center in the cloud—though it’s important for management to understand that such staffing costs do not completely vanish: in-house engineers, technicians, and cloud subject matter experts are still required to create, monitor, and support cloud infrastructure.
Another “cost” avoided when choosing the cloud is time. While it can take months to fill purchase orders, install support infrastructure, and stand-up in-house servers, cloud infrastructure can be set up in days or less, and can be configured to automatically scale with the workload or traffic. This saves money, since you aren’t over-provisioning, and also provides a good user experience. The ability to only pay for what you use, and horizontally scale as appropriate, makes the cloud very attractive for many businesses.
A data center hosted in the cloud has a fundamentally different capital and operational expense ratio to one hosted on premises. An on-premises data center requires enormous up-front capital costs, including building infrastructure, electrical and safety systems, and, of course, computer, networking, and storage hardware. On-premises data centers also require dedicated staff to manage the local hardware, an operational expense partially avoided when using cloud infrastructure. By reducing capital expenses in exchange for new operational expenses, an organization can reduce overall costs. The initial cost of setting up a data center in the cloud is much lower, and can scale to the precise level needed by the organization.