Getty Images

Automation: The key to FinOps success

Workload management and automation offers a way for FinOps teams to match supply to demand for cloud-based IT in an efficient way

Strong demand for cloud services continues to drive growth in IT spending this year, despite inflationary pressures, says Stephen Minton, vice-president of the data and analytics research group at IDC, in its Worldwide black book monthly analysis.

According to Minton, it’s just about the only tech category that is looking healthy at the moment – and with good reason. Cloud computing has fast become the backbone of modern IT infrastructure, but for all its good intentions, cloud also comes at a cost, and that cost is rising.

It makes for some difficult internal conversations between technology leaders and finance leaders. As a recent Nutanix Enterprise cloud index study found, cloud cost control ranks as a top IT management challenge for 85% of organisations. Similarly, a Flexera State of the cloud report found that managing cloud spend (82%) ranked higher than security concerns (79%). And a recent Cloud impact study from Aptum highlighted what it calls “cost leaks” and “runaway cloud costs”, suggesting hybrid or multicloud adoption has led to complexity.

As IDC says in a recent report, while cloud adoption has accelerated in the past couple of years, “cloud governance and control mechanisms haven’t kept pace”. As a result, it adds, “up to 30% of cloud spend is categorised as ‘waste’ spend”. This, says the firm, has helped nurture a shift towards cloud value, and central to this idea is financial operations, or FinOps.

Cost and culture

According to the FinOps Foundation, which describes FinOps as an “evolving cloud financial management discipline and cultural practice”, it is about bringing transparency and accountability to cloud spend. This should, in theory at least, enable organisations to manage cloud contracts and chargebacks better, but also find efficiencies in workload management.

The savings can be impressive. According to Christopher Squibb, FinOps lead for NHS Digital’s Cloud Centre of Excellence (CCoE), having a centralised view of all of NHS Digital’s cloud costs and usage has been transformational. It has already led to savings “in the region of several millions of pounds, approximately a 40% saving”, thanks to FinOps, he says.

Like many organisations, NHS Digital has embraced a cloud-first strategy, but this has its drawbacks. “It’s difficult to truly understand the cost impact of pursuing what is effectively a decentralised approach,” says Squibb. “With our previous on-premise technologies, we had a very clear view of upfront and ongoing costs. Adopting a cloud-first [approach] has removed that certainty.”

Read more about artificial intelligence in IT operations

  • Automation and AI are improving how the service delivery team at Leeds Building Society assesses the business impact of an IT service degradation.
  • While they might seem inherently different, both FinOps and DevOps encourage practices to improve business efficiency and success. Review the concepts and how they compare.

Squibb worked with UK reseller Softcat to implement a FinOps approach to “provide the visibility and data NHS Digital needed to simplify financial management, streamline cloud operations, and bolster security and compliance,” says Chris Redding, datacentre and cloud specialist at Softcat. However, it’s not something you can achieve overnight. Redding points out that the technology is only part of the solution. Integration and culture are key to ensuring the true value of a FinOps approach.

Culture is clearly a big issue. The FinOps Foundation’s recent State of FinOps research suggests over a third of organisations are struggling to empower engineers to take action on FinOps, despite its efficiency and cost-saving claims.

According to Harish Grama, global cloud practice leader at managed service provider Kyndryl, it’s about “creating a cost-conscious culture”. Grama was formerly CIO of cloud services for JP Morgan Chase & Co and spent three years running IBM’s public cloud.

In FinOps, he sees an opportunity to bring together, for the first time, financial, technical and business functions to drive financial accountability. But FinOps, he adds, “is not just a cost-saving tool, but a way of operating differently that enables employees across the business to work more effectively, leading to better optimisation of their spend”.

Automation focus

The FinOps Foundation illustrates this point. It has created frameworks to help organisations use appropriate tools to meet specific needs. A big area beyond purely measuring and optimising cloud spend is workload management automation. This focuses on “running resources only when they are needed and creating the mechanisms to automatically adjust what resources are running at any given time”, says the foundation.

This is intended to give FinOps teams the ability to match supply to demand. This means cloud usage can be optimised through the measurement of workload demand. The problem is that the level of automation required is far from being met, at the moment. The foundation points to a lack of automation maturity here – it claims that the majority of organisations (77.5%) are at a “crawl” stage, with just 3.2% running.

“Currently, the role AI [artificial intelligence] and automation plays in FinOps is nowhere near the level it should be,” says John Grubb, senior director of data and FinOps at “However, in the near future, AI involvement in FinOps will be considerable, and the process is likely to be entirely managed with the help of automation. This is a very hot market to be in, because the cost of goods sold and operational expense impacts of getting it right – or wrong – are potentially huge.”

Grubb sees automation as ultimately essential to get the most out of a FinOps approach. One of the concerns he has, at the moment at least, is that FinOps projects can be “incredibly data heavy” and “require a pretty deep knowledge of cloud architectures, financial terminology, and data modelling and querying, to be able to pull the job together in an impactful way”.

While he sees the overall goal as improving not just the operational efficiency of an organisation, but also the relationship between two historically distant departments within a typical technology business setup – finance and engineering – the lack of automation maturity remains a concern. Grubb concedes that “data and reporting needs will be improved in the near future, with LLM [large language model] and other forms of cognitive assistance becoming more operationalised”.

For Eamonn O’Neill, chief technology officer and co-founder of cloud managed service provider (MSP) Lemongrass, automation is a huge opportunity. It can, he says, be leveraged across a number of services, such as budget management. He says creating cost change estimates from automation code simplifies the generation of change requests for approval and ensures budgets change in line with the change request approvals.

O’Neill also suggests capacity optimisation is another good use of automation, where “self-healing” can “read” capacity needs and automatically re-allocate workloads, to minimise spend and avoid incidents of capacity running short.

“Service innovation is much harder to do without automation,” adds O’Neill. “Using automation to deploy and change a landscape makes it easier to switch from installed services to cloud-native services, which typically reduce costs and improve quality. While it is early days for this crossover of skills, AI and ML [machine learning] can much better predict capacity requirements over periodic cycles and, tied to the capacity optimisation point above, trigger changes automatically to better optimise spend.”

This is key. While cloud cost monitoring and reporting tools, such as VMware’s CloudHealth, Finout, Densify, Harness and Spot by NetApp, are readily available, the need to drive more automation is paramount. Cost monitoring, reporting and allocation is core, but it still requires time-consuming training and some manual inputs. This becomes more acute given that so many organisations are playing catch-up having invested heavily in cloud for the past few years with little or no monitoring on cost and allocation efficiency.

As McKinsey reveals in its recent report, The FinOps way: How to avoid the pitfalls to realizing cloud’s value, too many organisations are playing a “wait and see” game. “As many as half of the organisations we surveyed delayed establishing mature cloud financial management practices, such as granular visibility into spend, governance, forecasting and optimisation, until their annual cloud spend had reached $100m,” says the report.

So, where do organisations go from here?

Kyndryl’s Grama says: “Identify the current state of your organisation’s FinOps maturity and where you need or want to be. For instance, where is the spending happening and why? Create a new operating culture and model. Begin by asking, ‘How do I make things better?’. Deploy the relevant systems and adopt FinOps best practices into your business. Once the FinOps model is defined and adopted, continue to monitor and optimise the solution, ensuring that your organisation is also getting maximum value from its provider.

“Depending on the requirements of an organisation’s unique environment, users must continuously operate their FinOps model. It’s important to put emphasis on the importance of transparent visibility in creating awareness and driving a culture shift to operationalise FinOps for the longevity of the business.”

The whole point of the cloud is flexibility and speed. According to the FinOps Foundation, using FinOps means that technology, finance and business leaders can start to talk the same language and share the same processes, so that cloud spend and allocation can be managed more closely.

The idea is to invigorate and not undermine cloud use. But it can do so much more – at least, that’s the theory. Collaboration on data-driven spending decisions is a goal of most, but not all, organisations. FinOps, it seems, can be what you want it to be. It’s a state of mind, a culture, and the tools to make that culture more dynamic and automated are getting better all the time.

Read more on Managing servers and operating systems

Data Center
Data Management