Cloud adoption: A 2020 vision

Why is embracing the cloud so challenging?

I believe one of main reasons is that for the first time since the early 90s, almost all companies have simultaneously decided to transform their IT estate. The last big change on this scale was when PCs became cheap enough to be deployed on hundreds or thousands of desktops and client-server software was created. Nearly everyone decided at the same time to migrate from central mainframe systems to distributed platforms built on a variety of new software languages, creating the similar challenge of knowledge, experience and available resources.

Whilst it has been perceived to be of enormous impact, the subsequent introduction of the Web actually only caused a subset of enterprise applications to be rewritten, and this happened over a fairly long period of time. But of course in migrating now, the problem of legacy is far higher than it was 30 years ago, and the skills required to migrate or re-engineer for the cloud are far more complex and broad.

Take for example an article discussing client-server from 1996: “As technology sophistication increases linearly, the design skill requirements increase exponentially. Users are seduced by easy-to-use development tools. Today, anybody with a credit card can purchase visual development tools and create applications. When this is done outside the scope of the centralized IS department, these applications become support nightmares.” Brown observed: "A lot of those people doing those applications don't have any idea of what they're doing. They need the design skills.”

All this sounds pretty familiar 20 years on! History is repeating itself and many companies are repeating the same mistakes as before. In a 2018 report on the progress of cloud adoption, we highlighted some of the likely problems people would face with moving to cloud, and they still sound very familiar today:

• The evolving skill sets needed to deliver widescale technology estate transformation
• Lack of experienced people on cloud projects
• Changes in operating models (including internal billing and service management)
• Security and environment design
• Public cloud adoption framework and related operating model
• General public cloud (knowledge) resource capacity
• Where to obtain practical cloud knowledge from (cloud vendor, local consulting, offshore)
• Approaching regulators with their cloud based strategy
• Containerisation knowledge and delivery
• In-depth understanding of the security / compliance requirements of cloud matched to differing regulatory environments

So what is actually going on right now and where are people having issues? Let us start with the challenges and failures that we have seen. They broadly fall into the following categories I have tried to avoid mentioning the obvious ones):

Delays, delays and more delays

It is taking so long to build the foundations, that the underlying cloud technology has moved on in the meantime. Some teams are having to throw away what they have built, as by the time it is completed, it is already out of date. When you consider that cloud companies can introduce 3 or 4 major releases to their products per day, it is hardly surprising that internal teams are forever chasing their tails.

The challenge is bigger than you think

Teams are completely underestimating the size of the task to build out anything but the basics for cloud migration. Sometimes the teams are stating that this is a hugely resource-hungry endeavour, but are failing to convince their senior management teams to provide the resources. However even when they do get the resources needed, everyone seems to be ‘learning on the job’ and there is a real lack of experience. This has led to central teams trying to limit the workloads moving to cloud and the rise of huge amounts of Shadow IT. In a public cloud environment this can be a very risky approach.

Lack of knowledge in depth and breadth

Not only is the task a large one, but the equivalent knowledge of how to build physical datacentres has been built over literally decades of improvement, with hard-won skills becoming ever more niche. Think about this in a cloud environment – let’s take networking as an example – now you need deep skills in the following: WAN, LAN, firewalls, VPCs, network access, messaging, network costing, network analysis etc. It would be very hard to find those skills in just one person, yet this is just one area that a cloud datacentre requires. The knowledge of how to build a physical datacentre for many large companies is spread across tens, if not hundreds of people. Yet the size of the teams trying to build cloud infrastructure are normally a tiny percentage of that number.

Continued use of manual processes

The cloud promises a software defined datacentre, but almost all the people I talk to are still using manual provisioning (via the CSP console) and manual approvals. There is appetite to automate process and reduce the ‘path to production’, but it is a non-trivial task.

Lack of real experience to operate at scale

As pointed out earlier, most people are attempting this for the first time in their careers and are learning through trial and error. There are very few people around who have a track record of building the required foundational components and the automation required to contemplate large-scale cloud adoption or migration.

So whilst the benefits of large-scale cloud adoption are well known and continue to drive the desire to change, these are some of the very real problems that hinder such progress. However, I believe a practical solution to these problems can be achieved with a cloud adoption framework (CAF) utilising a ‘Datacenter as Code’ (DaC) solution. This builds on the Infrastructure as Code (IaC) concept; if IaC provides the ‘lego bricks’ then DaC is the completed model, following the detailed instructions. DaC is a living, running piece of software that automates the building and provisioning of all of your software-defined datacentre assets.

There is a real risk that as an industry of innovators, we are in the midst of repeating the same mistakes made in the mid-1990s. We certainly have similar constraints on accessing talented staff who have done this work before, and we are all going through the pain at more-or-less the same time.

I have faced these challenges many times working with some of the largest clients in the financial services industry; experiences that were at the core of our thinking about how to solve the problem together, for the benefit of all. A code-based open source initiative may provide the breakthrough the entire industry is crying out for, and I believe this approach can also be applied to many industries wishing to realise the benefits of rapid cloud adoption.

We don’t need to repeat the mistakes of the past; now is the time to approach cloud migration in a dynamic new way – with a clear 2020 vision of what a successful cloud-enabled future can hold.