Demystifying the private, public and hybrid clouds

The choices for adopting cloud services into your overall IT strategy are becoming more varied and there are several considerations in order to make the right choice for your business.

While many enterprises are familiar with both public and private cloud deployment, hybrid and multi-cloud models are a more recent innovation in the cloud computing space. A multi-cloud model is designed to scale with specific workload requirements and therefore demands a mix and match of cloud services (which are not connected or orchestrated) from different providers. A hybrid cloud however, is an orchestrated cloud architecture that integrates the on-premise private cloud with a third-party public cloud service.

Operationally, a multi-cloud strategy prevents vendor lock-in and avoids single points of failure and downtime since not every business function or application will have similar performance and security requirements. Hybrid clouds, however, integrate legacy systems to the public cloud environment to exchange/connect data with the on-premise private cloud. The goal here is to optimise the existing IT infrastructure while taking advantage of the public cloud scalability and performance.

In 2019, experts predict wide adoption of hybrid cloud adoption[i] as companies seek efficiency and budget optimisation. As enterprises continue to engage digital transformation initiatives, hybrid clouds will offer an attractive solution to complete this challenging transition for successful outcomes.

One of the biggest considerations in exploring any cloud deployment models is really about the level of control you expect in managing your cloud environment. Let us use the notion of "control" as the basis to help you determine the model that best fits your IT strategy.

With the private cloud’s build-and-operate model, you tend to have way more "creative design" control over the hardware and software components used compared to other models. From the Microsoft, VMware and Openstack stacks, to the wide array of hardware components in servers, storage and networking, there are numerous choice permutations, coupled with equally numerous licensing, pricing and configuration options. The biggest benefit of such an approach is that you have the deepest level of control to ensure that your private cloud is built with very specific support for the applications that you expect to run on this cloud, including addressing data residency and sovereignty compliance requirements. The downside of such an approach is however, that you will have to be deeply involved in every technology choice across your entire private cloud platform. And more importantly, you need to have deep technical resources and expertise in-house to help you make the right decisions in picking the right hardware and software solutions.

In adopting the public cloud such as Microsoft Azure and Amazon Web Services (AWS), you will have to depend heavily on the public cloud provider to design, build and continuously evolve or enhance their platforms as they deem fit. Computing resources are now presented to you as a utility service accessible through the public cloud providers’ portals/control panels or service application programming interfaces (APIs). You will have little to no control over the hardware and software choices as these are largely dictated by the public cloud provider. Your applications architecture starts to adapt and morph towards leveraging the public cloud providers' service APIs, and a more concerted effort will be required to re-architect your applications to be more "cloud-native". The biggest benefit of adapting to public clouds is that you are now able to leverage the highly competitive and rapidly progressing software innovations coming from these global public cloud providers. With new cloud services and enhancements on existing ones emerging almost on a weekly basis, you can focus on improving your business applications or platforms to support your core businesses rather than spend precious resources and time on lower level infrastructural concerns that your business stakeholders might never fully appreciate.

In the hybrid deployment model, you will see the benefits of both worlds mentioned above. Maybe you already have existing business applications and are exploring whether to "lift and shift" some or all of these applications to public clouds and "leave behind" some of these legacy applications. Or perhaps you are looking to re-factor or re-architect some of these applications to take full advantage of cloud-native features of public clouds. The "lift and shift" approach is never the most cost-efficient migration approach and the only time that such an approach makes sense is when you are hemorrhaging monies in maintaining your own physical infrastructure more than what you will incur on utilising public clouds.

The private cloud’s build-and-operate model again lends itself to a higher level of control and predictability of the costs involved in the build-out, ongoing consumption and operation of such an infrastructure. While you will have more control over your spending, your choices of flexing that spend becomes drastically limited as the investment is generally committed upfront, which is further compounded by the looming future of ongoing costs to ensure your hardware and software components are up-to-date. If your applications landscape is generally stable with predictable usage patterns, there are definitely more favourable cost containment reasons to be more heavily invested in your own private cloud infrastructure, which can tolerate longer technology refresh cycles that will help minimise the long term capital and operating expenses.

In the world of public clouds, the promise of flexibility and rapid scaling up or down of cloud services is very real. However, this extreme flexibility comes with a spend management headache if one is not careful. Imagine in your home right now, you might possibly have somewhere between 5 to 15 water taps. Some are turned off, some are on (dishwashing, showering, etc.) and some are timer-based (garden sprinkler systems). This is generally very manageable and wastage is easily controlled. Now imagine in an alternate reality world, additional water taps can be installed randomly at any time and anywhere, and suddenly you have 2,000 water taps to manage. You can now begin to see how managing this utility (i.e. water) can easily get out of control. Shadow IT and many different project teams in an enterprise can now utilise public clouds with the swipe of a credit card and very soon understanding and deciphering your public cloud spend can become a nightmare. This is very real and there are many companies today grappling with controlling their public cloud spend.

The hybrid cloud model starts to tamper the inflexibility of private cloud infrastructure spend with some flexibility afforded by public clouds. The considerations will then be focused on your applications' architecture and computing resources utilisation patterns. While there is no universal right answer, aside from the considerations mentioned above, re-factoring or re-architecting your applications to take full advantage of a hybrid cloud infrastructure might present you with the best of both worlds. What this means is that your applications' architecture becomes better distributed, and the internal components’ “placement” in these applications is based on what makes sense to situate in a private cloud infrastructure versus what makes sense to be designed into a public cloud infrastructure. Simply thinking in "enterprise, 3-tier architecture, monolithic" terms is no longer scalable and flexible in the new world culture of DevOps with continuous integration and continuous delivery.

Based on recent experiences with customers who are adapting to a public cloud deployment model, it is clear there is no one single universal answer for an organisation – in terms of whether the choice should be a private, public or hybrid cloud. Most organisations right now have islands of private and public clouds implementation but they might not be necessarily "integrated". However, there is indication that the hybrid cloud deployment model will evolve rapidly as IT professionals mature in their thinking on how to re-factor applications for a hybrid cloud landscape.

Singtel has the full list of services to help enterprises move off their legacy environment to the cloud. Contact us to find out more.

[i] https://www.rightscale.com/blog/cloud-industry-insights/cloud-computing-trends-2016-state-cloud-survey