Five lessons learned about cloud

AWS, Azure, Google Cloud and Alibaba Cloud posted growth of up to 40% in Q3 2020, despite the global crisis. Since the first lockdown, public cloud has attracted companies with its flexibility and guaranteed service during the pandemic. For example, this is the case of retailers which have found in the cloud a tool to meet the skyrocketing demand for online shopping. For other sectors, such as air transport or the hospitality industry, the promise of substantial savings and increased agility will continue to increase demand and support their cost-cutting strategy, to mitigate the economic consequences of the lockdowns.

However, for this promise to come true, it is essential to avoid some common cloud adoption pitfalls and joining the club of disappointed cloud users.

Conversion with no conviction

Many IT departments only decided to move to cloud to comply with senior management’s desire to modernize infrastructure or react to the spread of Shadow IT*. This purely defensive and half-hearted approach results in transformation plans that may be hesitant, incomplete and/or too slow. This does not address the needs of the business but makes organizations reluctant to adopt a comprehensive cloud strategy. They don’t dare switch to an entirely different system, decrease the size of data centers, or shut them down. Rail or air transport, where business is highly dependent on customer confidence, may be tempted by this choice as a risk reduction strategy.

Maintaining dual infrastructures not only increases operational costs but also increases complexity, resulting in the enterprise increasingly incapable of adopting digital and creating agile solutions which are critical for success. This slow uptake directly affected organizations employees during the first lockdown, with employees forced to adapt their work schedules according to data center resource availability, when it should have been the other way around.

"US e-commerce sales jump 37% in Q3 [2020] and nearly $1 in every $5 spent in the period came from online orders.” Digitalcommerce360

Data centers seek revenge

Even among those who believe in public cloud, old habits die hard. Throughout their careers, infrastructure managers have been risk averse when building data centers. In installations planned to last at least ten years and intended to host sensitive applications that are demanding in terms of capacity and security — they know that everything needs to be perfect from the very beginning.

Too often, this pattern is perpetuated in the cloud and immediately oversizes the architecture. This precaution is both unnecessary and expensive, since the cloud enables growth, progressive adaptation and even the ability to make mistakes (what a relief!). Rather than make long-term plans, it’s better to enable organic and iterative cloud infrastructure growth by making maximum use of cloud native tools that may be strengthened if needed.

It’s “open bar” for developers

However, this flexibility doesn’t mean that there’s no need for a solid framework! Heavily impacted companies must reinvent entire parts of their business in an emergency, for example hotels that want to switch to a coworking/teleworking service or airlines that propose cargo instead of passenger's flights. However, they may be tempted to open the floodgates too widely, fully focusing on innovation to survive. Developers love innovations, but a failure to set clear policies is equivalent to letting everyone pick what they want from a cloud provider’s comprehensive catalog. This would create an environment impossible to govern and operate due to its heterogeneity and complexity. Even if a portion of the work is assigned to cloud developers, standardization, rationalization and resource pooling are still required.

To define a framework, the simplest and most economical way is lift & shift or mapping existing resources in the cloud. Costs can be cut by up to 30%. Refactoring, which consists of restructuring an application to make better use of cloud’s native capabilities, offers greater benefits – from 40% up to 60% in the long run – but its costs (restructuring, infrastructure duplication, project timeline, etc.) should be taken into account. Therefore, it’s preferable to reap the benefits (closing of datacenter’s areas data storage, licenses, etc.) and then redevelop if necessary.

Code name: FinOps

As opposed to conventional financial management, which is proactive and based on forecasts and budgets, the cloud requires responsiveness. To optimize costs, it is vital to adjust to actual demands of the business ant not the demands of IT. This entails close collaboration between finance and technical teams to define dynamic, practical management rules that can be updated as needs change. The purpose of FinOps is to make this dialogue possible, specifically through consumption analysis and regular committee meetings. Although this key mechanism is simple to introduce, its implementation is often delayed. Since consumption has not been monitored from the start, the number of drifts increase and after a few months, hundreds of zombie instances are running but no-one knows what they do or why. This can eat into the expected benefits. A post-lockdown recovery strategy should include this dimension from the very early stages, since it is very difficult to take a backward step. Someone with a technical background and a good understanding of cloud provider invoicing methods should be the intermediary for purchasing divisions and business units. This may be a quick win for the retailers, which have a high-performance purchasing culture. They should easily be able to implement this approach, thus limiting the future excesses linked to the growth of online commerce.

An unfinished transformation

Due to resource distribution, cloud changes the very structure of the information system as well as the way users interact with it. However, very few companies have made their non-technical functions aware of the matter. There is much to gain when finance knows how to account for variable cloud costs, purchasing can manage subscription models, and business units understand the significant savings possibilities. Simply switching unnecessary equipment off at night or during weekends can save up to 60%, as can choosing 99% availability instead of 99.9% for non-sensitive applications. Any cloud strategy must consider precisely what the needs are, as well as an understanding that culture needs to be addressed as a key promoter or inhibitor in adopting cloud.

You may recognize in one of these cases that you or your company have experienced.

Then it's time to talk about an additional lesson. This may be the most important one. You have the opportunity to take advantage of the generalization of the cloud to reinforce the right to make mistakes in the corporate culture.

The right way to deploy the public cloud, is the one that will meet your expectations. Define them clearly, then experiment: “test and learn”. Don't be afraid to make mistakes. Because, the common element of disappointing transitions is that they have been carried out in fear of making bad choices that would prove irreversible. The public cloud frees you from this, all you have to do is find out how to get the most out of it!

*unregulated systems setup by users without the IT’s knowledge

By Mick Briggs, Portfolio manager at Atos

Posted: February 9th

Build your cloud forward future

Discover Atos OneCloud

Share this blog article


About Mick Briggs

Portfolio manager at Atos
Mick Briggs is a portfolio manager working to create initiative solutions for our customers. With over 25 years’ experience in Information Technology 10 of which has been working with cloud Mick has led Atos cloud solutions architects and engagements across the globe and is recognized an Atos senior expert in cloud.