Although a wide-scale move to a ‘cloud first’ strategy revolutionised the way businesses operate, after many years of cloud transformations, the most forward-thinking organisations are now taking the next great stride. They are developing hybrid cloud strategies to support everyday business operations while repatriating business-critical data-heavy initiatives back to on-premises systems.
One reason for this is that accelerated cloud migration fueled by the overnight move to working from home due to the global COVID-19 pandemic, combined with the growing complexity of navigating cloud usage and costs, has seen organisations in every sector battling to control their cloud spending and reduce waste.
These efforts are being hampered by a volatile economy, particularly in South Africa where the rand dollar exchange rate has soared from under R10 to the dollar in 2013, to nearly R19 to the dollar today. This is particularly true for companies using public cloud services such as AWS or Azure, as many organisations hosting workloads with hyperscalers are now finding themselves way over budget year after year.
If you add to this the always-growing volume of data being created today, as well as innovative, yet data-heavy workloads such as artificial intelligence (AI) and business intelligence (BI) that are being run in the cloud, you're left with entities across the board that desperately need to find another solution to balance the management of their cloud costs and technology budgets.
Unfortunately, the cloud-first strategies that skyrocketed a few years ago, have now reached the limit of their effectiveness, and in many cases, returns on investments are shrinking, setting off a major cloud backlash. Ubiquitous cloud adoption has seen a wide range of new challenges rise to the fore, such as out-of-control costs, increasing complexity, and vendor lock-in - in other words, cloud sprawl.
Lack of visibility
Many companies are finding it difficult to get their cloud costs under control, and a major cause of this is a lack of visibility into their cloud usage. Unfortunately, this lack of visibility has real implications for businesses’ bottom lines. As they struggle to get a concise view of their cloud costs and the efficiency of their cloud usage, they cannot hope to effectively manage their cloud spend and keep within budget. Cloud costs have been on the rise for years, and no business head can claim that a variable cost that keeps on accelerating can form a stable platform for the business.
Therefore, with cloud spending becoming one of the most expensive resources for a growing number of entities, business and finance leaders are looking for strategies and tools to optimise their cloud spend, eliminate waste, and gain a central view of their cloud cost management.
One of the cloud’s major benefits is its lack of ceiling when it comes to scalability, but this is a double-edged sword. Businesses that once worried about not being data rich are becoming data drunk, and unless they can bring at least part of their data estates back on-premise, bringing these costs under control will be impossible, and will negatively affect their bottom line.
The sheer number of workloads in the cloud is causing cloud expenses to soar out of control. Companies are now running not only critical compute workloads but enormous storage volumes in the cloud, as well as data-hungry applications such as machine learning (ML), BI, and AI in the cloud, all of which need dozens or even hundreds of GPUs and terabytes or even petabytes of data. This results in heavy workloads being crunched in the background, leading to enterprises using a lot more compute and storage than they budgeted for.
The question then is, if the cloud is unmanageable and too costly, how should organisations proceed? Should they take all their workloads off AWS and Microsoft Azure and move everything back in-house? In short, no. While moving back on-prem might make economic sense for some businesses, it is too big and complex a task for others. While some cloud expenses are prohibitive, many entities are still reaping tremendous value from having certain applications in the public cloud.
A move to a hybrid cloud
A hybrid solution is the answer, as it enables companies to regain control of cloud costs. For example, Microsoft 365 runs in the cloud, it was designed for the cloud and should stay in the cloud. It is a fixed cost per user each month, so it makes sense to keep it there. However, companies should look to bring their AI, BI, ML, and even MySQL back on-prem to control these variable costs. If these workloads are run on-prem, it doesn’t matter if a user needs to run a big program overnight, as they won’t be charged extra for it.
It’s important to stress that no one is suggesting that companies rebuild their data centres, especially in South Africa where the electricity supply is no longer reliable and heavily interrupted by load shedding and cable theft. They can col-locate at any of the many dedicated Data Centers in South Africa, for example Teraco or Vantage, and host their own servers there. Or they can talk to a trusted provider who can provide them with hardware-as-a-service or infrastructure-as-a-service within a shared or hosted environment. In this way, they are not paying for all the compute, storage, and networking, and costs are brought back under their control.
The bottom line is that applications that require high-performance computing are better and more cost-effectively run on-premise, while the cloud is perfect for cloud native applications as well as less-demanding applications.