Private Cloud vs Public, Hybrid or Multi Clouds
The data center continues to morph – from on premise to private and public clouds, hybrid clouds and multiple clouds. IT can no longer govern which options its users are choosing, but is still responsible for meeting the company’s business demands, ensuring performance, governance, security and more. With so many organizations pursuing a cloud first strategy, the discussion is no longer should it be cloud, but instead focus on the cloud technology options that best realize enterprise requirements or specific use cases. What is driving businesses to the cloud? While a number of factors come into play, the primary concerns center on budget, availability and complexity.
Public clouds potentially offer lower costs with no CAPEX investment in infrastructure, but as a deployment scales up, costs may skyrocket. Security and compliance are other common reasons that organizations may hesitate to move certain resources to the public cloud. However only 20% of workloads are on public cloud IaaS, meaning that the majority of cloud resources are consumed in one of the private cloud configurations. As Thomas Bittman explains “public cloud VMs are much more likely to be used for horizontally-scalable, cloud-friendly, short-term instances. Network performance on public cloud resources may suffer, as latency may increase with traffic moving within the cloud and between company sites and the cloud. If performance is a prime consideration, select a cloud service provider with many data center sites.
Private cloud options range from semi-private, virtual-private, or fully-private capabilities. Some of the advantages of private cloud include greater security and control. Upfront, the cost may be higher, due to utilization of in-house IT, requiring purchase of software and hardware resources. The workloads that tend to be placed on private clouds are usually more traditional, with vertically-scalable, long-term instances.
Hybrid cloud provides a ‘best of both’ option that gives enterprises the flexibility to choose where to place a workload. Usually mission-critical systems such as databases remain in-house, whereas those that require intensive processing such as big data analytics, benefit from the scalability and power of the public cloud. Some applications may require modification to adapt to the cloud environment where instances are spun up and down, therefore significant resiliency testing is required before migrating apps to cloud environments.
Cloud bursting (scaling to public cloud on demand) provides increased resources for high or peak load requirements, scaling back when the extra compute capacity is no longer needed, providing a distinct saving on infrastructure costs.
Taking hybrid to the next level, many enterprises have chosen to distribute their workloads among different cloud providers leading to a ‘logical cloud,’ meshing resources from a number of public clouds and on-premises infrastructure. With a multi-cloud approach, the most appropriate SaaS, PaaS and IaaS providers can be selected. Platforms, services and applications can be chosen according to the requirements of each workload. For example, apps built using Microsoft .NET are best off in a Microsoft Azure cloud, while the Google App Engine is a good choice for apps built in Java, Python or Spring. Smart workload prioritization and allocation enables highly-efficient resource utilization.
Since each of the public cloud providers utilize different APIs, coordinating these workloads may require some recoding. Porting cloud services is easier between providers that support the same API, such as Rackspace and HPE Helion that both use OpenStack.
However, managing multi-cloud environments demands some overhead, often requiring an MSP with specific expertise. This effort usually is offset by the benefits of multi-cloud which include avoidance of vendor lock-in, high availability and mitigation of service disruption as resources are distributed among service providers and locations.