By Peter Lunk, Chief Marketing Officer, CloudSphere, Abhijit Joshi, CEO, CloudHedge, Stephanie Hung, Senior Vice President / Head, Cloud Business, ST Engineering, Abhishek Pradhan, Vice President / Head, Cloud Technology, ST Engineering
Like clockwork, the IT department of an e-commerce company fires up an array of applications to manage daily operations. These applications are spread over multiple clouds, each designed to optimise different activities.
Fast-moving actions such as cataloguing and payment are hosted on clouds with high upload speed, while marketing content is stored on clouds that provide faster video encoding.
Adopting a multi-cloud strategy allows companies to leverage different public and private cloud platforms to achieve various benefits. These include greater efficiency and agility, avoiding vendor lock-in and ensuring competitive pricing.
As with all new plans, companies face tricky implementation processes such as cloud integration difficulties and potential overspending. Fortunately, there are steps to tackle these obstacles and build a resilient multi-cloud cloud infrastructure.
Typically, companies with one preferred cloud provider tailor their entire operations around it. But with more options such as Amazon Web Services, Microsoft Azure or Google Cloud Platform today, a multi-cloud strategy presents a world of possibilities.
Avoid putting all your eggs into one basket. The main advantage of a multi-cloud approach is that it prevents vendor lock-in, where organisations are beholden to one provider and have little leeway in pricing and flexibility.
Today, a wide variety of providers has created a competitive market. Each platform offers different advantages and cloud users can compare prices and terms of service before selecting the ones that best suit their needs.
Having alternatives allows companies to match their operations to a specific platform to optimise that part of their business. Users need to constantly think about which cloud should be used for every new application deployed, instead of sticking to a default one.
Another perk of a multi-cloud strategy involves risk management. For example, if a provider is compromised, companies can swiftly port their operations to another platform or back up their data in a private cloud.
However, multi-cloud adoption is easier said than done. When people start out with a single cloud, they often use proprietary tools provided by the vendor and get used to managing the cloud in a certain way.
Upon switching, they will run into difficulties such as not being able to run reports, look at costs or manage security as usual. The different tool sets of multiple clouds add to the complexity and integration issues.
In the past, cloud users also had to design applications to fit a different provider when switching platforms. This creates a “blind spot”, where there is a tendency to design applications to fit multi-cloud platforms, which increases costs.
Frameworks are also necessary to govern cloud usage across the company. Without visibility or guidance, costs may arise from incidents such as cloud sprawl, which is the unnecessary overuse of a company’s cloud resources.
For example, if developers run a new application on a cloud platform and forget to remove it when no longer needed, it will drive up business costs as public cloud providers are paid for the amount of bandwidth used per month.
In addition, migrating existing legacy systems is challenging and comes at a high cost given the huge amount of data and technical skills involved – although we seem to be at the tail end of the issue.
Optimising the multi-cloud
While third-party tools can be used to solve integration issues, users still have to manually customise them for each cloud provider. A better solution would be to use Kubernetes, an orchestration engine that automates container provisioning, networking and load balancing.
Containers are popular in hybrid or multi-cloud environments as they make it easy to shift workloads from on-premise systems to a public cloud by packaging codes to run applications faster. Kubernetes then manages these clusters uniformly across the different cloud platforms.
This can be governed by a framework, which gives companies a blueprint to manage their multi-cloud strategy and workloads while providing transparency across all platforms.
Companies should also look to optimise vendor offerings such as reserved instances – a billing discount in return for committing to a specified level of usage – and savings plans.
Companies should use software with pre-installed policy guidelines to help with this selection process. For example, users who want 60 per cent of their operations running on reserved instances and the rest bought on-demand can run that through a software, which will find the best deal. Let automation deal with multi-cloud complexities.
Lastly, match your applications to different clouds based on the provider’s setup. This cuts down on the cost of designing applications to work across many clouds and provides more choices for suitability in the multi-cloud environment.
To stay agile, companies should plan ahead for a multi-cloud environment. It may take two or three years, but with the infrastructure and framework in place, they can eventually realise their multi-cloud vision.