From Edge to Cloud: Understanding the Infrastructure Delivery Spectrum

In the not too distant past, the infrastructure world was a fairly simple place. There were servers, storage systems and networking equipment. The combination of these three raw materials provided the foundation for many applications and companies.

Over time, the demand and requirements placed on the underlying infrastructure changed while the sheer amount of resources increased too. One would think that technology would get simpler over time. Contrary to conventional wisdom, enterprises are facing an increasingly complicated situation when it comes to infrastructure. On one hand, the underlying technology is getting simpler to manage. However, on the other hand, the number of infrastructure deliver methods are increasing. This latter point is the key driver to increased complexity and complicated infrastructure management challenges.

While some may claim that enterprises only need one or two models or state that models are interchangeable, reality is quite different. Each delivery model uniquely offers a different set of characteristics. As such, no two models are the same nor interchangeable.

NOT ALL ROADS END AT PUBLIC CLOUD

Companies are looking for new ways to consume technology infrastructure without the limitations of traditional silos. One of those new consumption models is public cloud. Is public cloud the replacement to all traditional silos of corporate infrastructure? There is a common narrative that suggests that all applications (workloads) will eventually end up running on public cloud. While that may seem plausible in some circles, reality suggests something very different for the enterprise.

Hosted private cloud and public cloud infrastructure services did open up new capability for enterprises. Almost all hosted private cloud services are focused on support for VM-based workloads versus truly cloud-native workloads. Public cloud presents the ability to host either VM-based workloads or cloud-native workloads on the same infrastructure. That is not to say that public cloud is better than hosted private cloud. There are differences between the two in terms of dedicated infrastructure, cost models, security paradigms and architectures.

NEW ON-PREMISES MODELS

While hosted and public cloud models offer new opportunities for enterprises, they still leave gaps between what the enterprise needed, wanted and could use. Enterprises continue to strive for simplification of their infrastructure foundation without having to go all the way to public cloud. Keep in mind that there are large number of enterprise workloads that will never (or can never) be refactored for cloud-based infrastructure. The advent of hyperconverged and composable infrastructure present new ways to simplify the infrastructure without going all the way to cloud.

BEST OF BOTH CLOUD WORLDS

Even those workloads that could be refactored for cloud may still run into challenges with public cloud infrastructure. There are a myriad of reasons for this that range from cultural norms and inertia to regulatory and compliance requirements.

Private cloud presents a middle ground that addresses many of these challenges. Private cloud solutions that run on-premises offer similar functionality as that found in public cloud except that it runs within the corporate data center. It is important to keep in mind that not all public cloud functions are supported by private cloud solutions. Nor should they be. There are services (such as artificial intelligence, machine learning, analytics, etc) that are better suited for public cloud than private cloud.

THE NEW CORPORATE DATA CENTER

Beyond cloud, the corporate data center is changing too. In past posts, I wrote about the ‘Death of the Data Center’ and the demise of the traditional corporate data center. This is still true today but takes on a slightly different form. The need for the monolithic traditional corporate data center is going away and being replaced by smaller, and more numerous, data centers that are purpose built.

There are two core reasons driving this data center shift: 1) cloud and 2) edge and Internet of Things. As mentioned above, there are enterprise workloads that will not move to cloud. At the same time, the rise in edge and internet of things applications are driving the need for a multi-tiered infrastructure approach.

Enterprises are already starting to move to a new infrastructure model where smaller, purpose-built data centers support the edge and internet of things. The monolithic corporate data center is being replaced by a smaller mid-tier facility that supports a combination of legacy traditional silos, hyperconverged, composable and/or private cloud infrastructure. This is often used in conjunction with public cloud infrastructure.

Can this multi-tiered infrastructure architecture be replaced by public cloud? In a word, no. Many applications are sensitive to latency which is driven by speed of light limitations. Applications that rely on a closer data center often leverage mid-tier applications and public cloud applications for more sophisticated functions such as machine learning.

UNDERSTANDING THE SAAS BYPASS

Many enterprises struggle to refactor legacy applications for a myriad of reasons. Ironically, even if an application has a positive ROI that supports refactoring, it still may present a challenge for the enterprise to do so.

One approach that enterprises are taking is the ‘SaaS Bypass’. Instead of refactoring an application, the enterprise is simply replacing it with a SaaS-based alternative. This approach creates a clever solution to those applications that are not strategic to the business.

LOOKING BEYOND INFRASTRUCTURE

The key to solving this complicated problem lies with management tools. Today’s infrastructure management tools automate many of the manual processes through three core opportunities:

1. Speed: By automating the processes, it provides quicker response to changing demands and issues.
2. Error Reduction: It is well known that humans make mistakes. By automating processes, it significantly reduces the number of human-introduced errors to infrastructure configuration and management.
3. Integration: Modern management tools provide automation that spans across the classic server, storage and network technologies. It significantly reduces the need for technology specialization within these three core pillars.

The management tools are only one part of the change. Cultural norms, intertia and application modernization must all be addressed to successfully leverage the opportunities that the infrastructure delivery spectrum presents.