Knowing when to move to the cloud

Cloud adoption continues to increase rapidly, buoyed by advantages such as its scalability and the sheer breadth of capabilities that can be encapsulated in versatile cloud services.

One reason that enterprises are attracted to the cloud is its flexibility. With the ability to rapidly deploy infrastructure through a web console or prewritten script, enterprises can react quickly to new business requirements by rolling out new infrastructure in minutes or hours instead of days and weeks with zero capital expenditure, including the option to backtrack and to shut them down on a whim.

Buoyed by the growing maturity of cloud capabilities, it is hence unsurprising that complex multi-cloud and hybrid cloud deployments are expected to dominate the enterprise¹. Indeed, more than two in three (69%) respondents from a study² by analyst firm 451 Research said they planned to have some type of multi-cloud environment by 2019.

There are advantages that multi-cloud offers that are not found with a monolithic cloud deployment. Foremost among them is the ability to mitigate the risk of data loss or downtime stemming from localised cloud failures or outages. By not putting all the eggs in one basket, enterprises are hence able to increase the reliability of their cloud infrastructure at essentially no added cost.

Another consideration of a multi-cloud deployment is how some clouds are better suited than others at specific tasks. For instance, a software-as-a-cloud (SaaS) solution may offer better cost efficiency without the hassle and manpower overheads of manually deploying and configuring the same system on a preferred infrastructure-as-a-service (IaaS) cloud.

Yet despite the cloud’s many tangible advantages, not all systems work well in the cloud. Some workloads may not be suitable at all, while others may require extensive rewrites or rearchitecting to function smoothly in the cloud. Finally, there may be other limitations due to regulatory requirements or hardware performance that may prevent certain types of services from being moved to the cloud. So how does one decide?

The key is not to treat the cloud as the only destination for IT deployments, but to see it as a platform that works exceptionally well for certain types of applications, and poorly for others. For a start, workloads that work best in the cloud tend to revolve around mundane services such as email servers or cloud-based collaboration tools. Servers that are already virtualised or databases supported by the public cloud provider are also good candidates for deployment in the cloud, as with temporary staging or test environments.

What about deployments that are better left on-premises? Some of them include those that fall into the following categories such as high-performance systems, legacy services, latency-sensitive applications, and organisations in certain regulated industries.

We outline some of them below:

High-performance systems: While cloud infrastructure can offer vast amount of processing capability, they can substantially increase operational costs. There is also no point making a switch if the requisite high-performance systems are already in place.

Legacy services: Systems that are better off in data centres include legacy mainframes or client-server apps that were originally designed for on-premises deployment. While these could theoretically be ported over, they typically require costly code rewrites and testing.

Legacy-sensitive applications: Latency sensitive applications or services that are extremely bandwidth-heavy will likely not work well in the cloud or incur prohibitive bandwidth cost if deployed there.

For mature enterprises with a plethora of production IT systems and services, it may not be immediately obvious what should be migrated to the cloud and what should stay on-premises. The onus rests on the enterprise to make a proper assessment – by working with external cloud experts if necessary, based on a variety of considerations such as cost, maintenance, control and compliance with regulatory requirements for their industry.

It is vital that costs such as the specialised work and man hours required to migrate a non-cloud application to the cloud not be overlooked in this assessment, including applicable software licensing fees. Given how IaaS platforms typically operate by core count, this means that applications or solutions with rigid licensing terms tied to solely to the processor count could end up costing substantially more.

Ultimately, enterprises must understand that the shape and form of their multi-cloud or hybrid cloud infrastructure will not be the same as their competitor’s but will be a unique deployment shaped by their specific requirements and prior IT investments.

Speak to us to find out more.

¹ https://www.information-age.com/rise-multi-cloud-data-controllers-123466502/

² https://www.information-age.com/multi-cloudhybrid-environment-dominate-enterprise-123469737/