Subscribe
About

Hyperscale showdown

From all-in adoption to deliberate deployment, is it time to rethink how the cloud works for businesses?
By Tiana Cline, Contributor
Johannesburg, 07 Nov 2024
Chris Erasmus, AWS South Africa
Chris Erasmus, AWS South Africa

Hyperscale cloud has become an intrinsic part of the digital economy. Whether it’s people browsing the web, making purchases, or businesses managing data storage, hyperscale cloud is embedded in nearly every layer of modern infrastructure. Cloud, it seems, has become a business necessity. A recent McKinsey survey of around 50 African business leaders showed that 45% of their workloads are in public cloud, which is on par or ahead of the rates of adoption in North America and China.

Locally, reports show that South Africa’s cloud market will reach over R113 billion by 2028. All of the big players – Amazon Web Services (AWS), Microsoft Azure, Google Cloud Platform (GCP), Oracle Cloud and Huawei Cloud – have local availability zones. Alibaba also has a presence, through its partnership with BCX.

David Brown, vice president of AWS Compute and Networking Services, explains that these zones are how hyperscale cloud providers can quickly deploy infrastructure close to customers. “The important thing is not really the number of datacentres; if a customer needs to launch an instance or build a new application, or the business needs to scale, can they get that support?” asks Brown

“Hyperscale cloud adoption is accelerating.”

Chris Erasmus, AWS South Africa

Absa, Capitec, Standard Bank and TymeBank have each migrated a significant portion of their cloud infrastructure to AWS. Although Nedbank has a multicloud strategy, the majority of its cloud-based architecture is with Azure. First National Bank (FNB) uses AWS, Azure and GCP.

While it’s not clear which hyperscaler has the lion’s share of the local market (the companies themselves don’t report this), AWS is ahead in every region in the world except China.

CLOUD CORE

Cloud services are divided into three main categories: SaaS (Software-as-a-Service), IaaS (Infrastructure-as-a-Service) and PaaS (Platform-as-a-Service). Salesforce and Microsoft are currently leading the SaaS market, integrating AI to improve customer relationship management and productivity tools. In IaaS, AWS, Microsoft Azure and Google Cloud dominate with high-performance computing and AI-powered solutions. AWS focuses on GenAI through its Bedrock platform, while Azure offers hybrid capabilities, like Azure Arc, and Google promotes sustainability through energyefficient, AI-optimised datacentres. With PaaS, these giants provide developers with tools like Kubernetes and serverless computing to streamline app development and deployment.

According to Synergy Research Group, AWS is leading the global cloud infrastructure market with 32% market share, followed by Microsoft at 23%, Google at 12% and other players at less than 5% each. Even though the three US tech giants dominate cloud adoption worldwide, the Chinese hyperscale cloud market tells a different story. In China, the top three providers are local, with Alibaba Cloud leading the market, followed by Tencent Cloud, China Telecom Cloud and AWS China. (AWS operates through two partners, Sinnet in Beijing and NWCD in the Ningxia region.)

With the US government restricting US-made AI chips like Nvidia NPUs that power large language models from being exported to China, using AWS advanced AI models in the cloud is not a violation of US regulations. GCP is not available in China, and Azure is operated by 21Vianet, a physically separated instance of cloud services. China is an important market to watch, and while it is some way behind the US, its market is led by Chinese companies. It’s also constructing the world’s first commercial underwater datacentre off the coast of Hainan Island.

Deliberate innovation

As global hyperscale cloud providers continue to invest in South Africa, the market is maturing. Chris Erasmus, AWS South Africa’s country manager, says local cloud adoption reaccelerated in 2024. “The market is thriving – customers want to innovate.”

In the past, says Erasmus, businesses would start using cloud without being entirely sure of the benefits; the value was an afterthought, or, put another way, innovation for the sake of innovation. “Now we’re finding that many organisations are being a lot more deliberate about how they adopt, the pattern in which they adopt and the outcomes they want to enable,” he says. “They know the value and want to leverage it accordingly.”

Whether a business chooses AWS, GCP, Azure or some kind of hybrid or multicloud deployment, hyperscale cloud is complex. This is unlikely to change; there’s a dizzying number of as-a-service offerings available, with more on the way. As new services expand, so do the layers of complexity in managing infrastructure, security and costs. Erasmus laughs and compares AWS to Disneyland. “If you look at the technology and all the services and the pace at which we innovate and provide features, our customers can very easily get lost.”

He says there are many cloud training opportunities available, some of which are free. The challenge is applying those skills in a way that makes the most sense for an organisation, especially in a multicloud environment. “Being quite narrow and very specific around the skills that you have and the things that you do is much more meaningful than trying to understand the full stack,” he says. “Trying to figure out what you’re doing across multiple technology providers is incredibly difficult.” He recommends businesses take advantage of the specialists that hyperscale cloud providers have rather than trying to have all of the skills in-house.

The tools of the trade

Another complexity in the hyperscale cloud deployment puzzle is security. As businesses scale, traditional perimeter-based security approaches are insufficient and can leave gaps that can be exploited by increasingly sophisticated cyber threats. “A fragmented security landscape with overlapping vendors and solutions often results in security sprawl, making it difficult to maintain a cohesive defence,” says Ralph Berndt, director of sales and marketing at inq. South Africa. With multicloud, maintaining control over sprawling estates requires consistent security practices across different platforms. Each hyperscale cloud provider has unique security configurations, requiring administrators to dedicate time and effort to managing multiple systems, which can lead to errors. “Hyperscale environments amplify the risks of misconfigurations, which can result in data breaches or system vulnerabilities,” says Berndt. He suggests implementing multilayered security strategies, incorporating AI-driven security tools like endpoint detection and response (EDR) and network traffic analysis (NTA) for comprehensive visibility and threat detection.

“Hyperscale environments amplify the risks of misconfigurations, which can result in data breaches or system vulnerabilities.”

Ralph Berndt, inq. South Africa

AI is also transforming how businesses see and adopt hyperscale cloud. The excitement around generative AI, in particular, has many organisations realising that they must first modernise their databases in order to leverage AI effectively. AI’s pull has not only created a surge in cloud consumption, it has resulted in a datacentre building boom, with CBRE Research reporting that AI workloads have datacentre vacancies at an all-time low. “The entry point to generative AI has become an ah-ha moment for cloud,” says Erasmus. But for businesses looking to build more than a conversational chatbot, Erasmus says they’ll have to reconsider how their data is managed and architected, their staff strategies and operational models. “They need to get core systems and capabilities in place to move to the cloud and enable AI,” he says. “And there’s no slow down, at all. Hyperscale cloud adoption is accelerating.”

Bases on IaaS, PaaS and hosted private cloud revenues in Q2 2024. Source: Synergy Research Group
Bases on IaaS, PaaS and hosted private cloud revenues in Q2 2024. Source: Synergy Research Group

THE CLOUD EXIT CONVERSATION

Many CTOs and CIOs said they were “all-in” on hyperscale cloud during the 2010s. Everyone wanted to move from a capex to opex model and whole workloads were lifted and shifted to the cloud. Two decades later, companies are rethinking their strategies, but that doesn’t necessarily mean they’re leaving cloud entirely. A study by Citrix revealed that 42% of organisations are considering or have already moved at least half of their cloud-based workloads back to on-premises infrastructure, mainly due to unexpected costs. Cloud exit isn’t always about abandoning the cloud entirely, but, rather, about rethinking how to use it more effectively. What many businesses are realising is that exiting the cloud can be just as complicated as cloud adoption.

David Linthicum, Deloitte’s chief cloud strategy officer, says that when a business doesn’t get its ROI from cloud, it’s most probably due to inadequate planning, mismanagement of cloud resources or a lack of understanding about the cloud’s shared responsibility model. In other words, it’s not always about the cloud – it’s you. “Many businesses have felt the consequences of too quickly migrating to the public cloud without a well defined strategy in place,” says President Ntuli, managing director for HPE South Africa. “Similarly, there is a danger for CIOs in getting caught up in the growing trend towards cloud repatriation without first developing a robust cloud migration plan.” Ntuli says IT teams need to do some thinking around why they’re shifting workloads.

“In the end, your decision-making should come down to optimising and streamlining resources.”

* Article first published on brainstorm.itweb.co.za

Share