A cloud server is a hosted, and typically virtual, compute server that is accessed by users over a network. Cloud servers are intended to provide the same functions, support the same operating systems (OSes) and applications, and offer performance characteristics similar to traditional physical servers that run in a local data center. Cloud servers are often referred to as virtual servers, virtual private servers or virtual platforms.
Cloud server types and features
An enterprise can choose from several types of cloud servers. Three primary models include:
Public cloud servers: The most common expression of a cloud server is a virtual machine (VM) -- or compute "instance" -- that a public cloud provider hosts on its own infrastructure, and delivers to users across the internet using a web-based interface or console. This model is broadly known as infrastructure as a service (IaaS). Common examples of cloud servers include Amazon Elastic Compute Cloud instances, Azure instances and Google Compute Engine instances.
Private cloud servers: A cloud server may also be a compute instance within an on-premises private cloud. In this case, an enterprise delivers the cloud server to internal users across a local area network, and, in some cases, also to external users across the internet. The primary difference between a hosted public cloud server and a private cloud server is that the latter exists within an organization's own infrastructure, where a public cloud server is owned and operated outside of the organization.
Dedicated cloud servers: In addition to virtual cloud servers, cloud providers can also supply physical cloud servers, also known as bare-metal servers, which essentially dedicate a cloud provider's physical server to a user. These dedicated cloud servers – also called dedicated instances -- are typically used when an organization must deploy a custom virtualization layer, or mitigate the performance and security concerns that often accompany a multi-tenant cloud server.
Cloud servers are available in a wide array of compute options, with varying amounts of processors and memory resources. This enables a user to select an instance type that best fits the needs of a specific workload. For example, a smaller Amazon EC2 instance might offer one virtual CPU and 2 GB of memory, while a larger Amazon EC2 instance provides 96 virtual CPUs and 384 GB of memory. In addition, it is possible to find cloud server instances that are tailored to unique workload requirements, such as compute-optimized instances that include more processors relative to the amount of memory.
While it's common for traditional physical servers to include some storage, most public cloud servers do not include storage resources. Instead, cloud providers typically offer storage as a separate cloud service, such as Amazon Simple Storage Service and Google Cloud Storage. A user provisions and associates storage instances with cloud servers to hold content, such as VM images and application data.
Benefits and drawbacks of cloud servers
The choice to use a cloud server will depend on the needs of the organization and its specific application and workload requirements. Some potential benefits and drawbacks include:
Ease of use: One of the biggest benefits of cloud servers is that a user can provision them in a matter of minutes. With a public cloud server, an organization does not need to worry about server installation, maintenance or other tasks that come with ownership of a physical server.
Globalization: Public cloud servers can "globalize" workloads. With a traditional centralized data center, users can still access workloads globally, but network latency and disruptions can reduce performance for geographically distant users. By hosting duplicate instances of a workload in different global regions, users can benefit from faster and often more reliable access.
Cost: Public cloud servers follow a pay-as-you-go pricing model. Compared to a traditional physical server, this can save an organization money, particularly for workloads that only need to run temporarily or are used infrequently. Cloud servers are often used in such temporary use cases, such as software development and testing, as well as where high scalability is important. However, depending on the amount of use, the long-term and full-time cost of cloud servers can become more expensive than owning the server outright. In addition, regulatory obligations and corporate governance standards may prohibit organizations from using cloud servers and storing data in different geographic regions.
Performance: Because cloud severs are typically multi-tenant environments, and a user has no direct control over those servers' physical location, a VM may be adversely impacted by excessive storage or network demands of other cloud servers on the same hardware. This is often referred to as the "noisy neighbor" issue. Dedicated or bare-metal cloud servers can help a user avoid this problem.
Outages and resilience: Cloud servers are subject to periodic and unpredictable service outages, usually due to a fault within the provider's environment or an unexpected network disruption. For this reason, and because a user has no control over a cloud provider's infrastructure, some organizations choose to keep mission-critical workloads within their local data center rather than the public cloud. Also, there is no inherent high availability or redundancy in public clouds. Users that require greater availability for a workload must deliberately architect that availability into the workload.