AI and cloud increases need for workload management, say tech leaders

(Image by Tumisu from Pixabay )

Workload analysis and management never go away. The major business pressures of today are increasing the need for business technology leaders to concentrate on the economics of cloud workload management. A need to cut costs in the face of inflation and reduce CO2 emissions created by technology, while also allowing the experimentation and adoption of artificial intelligence (AI), ensures workload management is a high CIO priority. 

With technology now central to the operations and opportunities of a business, cloud workload management is a business practice, not a technology task. As a result, cloud workload management is forming a part of businesses meeting their environmental, social governance (ESG) commitments. Paul Petty, Group IT Platforms and Infrastructure Lead with Laing O' Rourke, the construction firm, says: 

ESG is one of our driving factors in terms of what we do. In IT, we are not big enough to make a dent in the carbon output of a construction project, but it can make a difference in how we do more with less. So we have utilization initiatives for our cloud and public cloud usage.

Customers, whether firms selecting a construction partner or consumers on the high street, expect organizations to be doing everything in their power to reduce the amount of CO2 emitted. IT is where some early reductions can be achieved through optimizing workloads, as Michael Allen, Head of Infrastructure at shoe retailer Clarks, says: 

We have switched off two data rooms, and the power saving is fairly significant.

Energy costs have risen up the business agenda since Russia invaded Ukraine, but the technology skills shortage is also adding to the cost base of business. Over half (54%) of technology leaders told the recent Digital Leadership Report that the skills shortage is preventing them from meeting their organization's change needs. For CTO Nick Ovenden at insurance firm Markerstudy, managing cloud workloads more efficiently has led to the redeployment of existing technology staff:

The people we have can now service more, and our people are repurposed to more useful roles.

It's not all about cost; cloud workload management has to be considered in a business priority context, as Markerstudy CIO Adam Miller says: 

It is a complex discussion on where to place workloads. So you need to evaluate on a use case basis of what the growth and expansion for the business will be, so it has to be based on hard facts.

Markerstudy has a team focused on overall spending on technology. An added complexity for Miller is that Markerstudy is a highly acquisitive business, so new workloads are constantly arriving as mergers and acquisitions are completed. He says: 

With acquisitions, you get a view through the due-diligence process, but when you get the keys, then you really see what is going on.

As a result, his team has a template for reducing the application landscape to keep workloads at an efficient level. Allen at Clarks agrees: 

It is about the right place for the right application. So, it is not just about cost; it is about agility. For example, cloud makes a lot of sense for the e-commerce platform, but not always for back-end applications.

Cloud computing has delivered that agility, but as Dunhumby CTO Ben Burdsall told diginomica recently: 

If you have a business problem to solve, there are now hundreds of ways to solve it due to the rich number of services on the cloud. That is where the challenge is, as the cost profiles are so different.

It is for this reason that cloud vendors, including AWS, Microsoft and Nutanix, are investing in teams of cloud economists. Steen Dalgas of Nutanix explained the role recently: 

We promised they're going to get these outcomes and this business case, and we'll go back and do the business case three years later. We'll rerun those numbers, and we'll ask, `Did we deliver what we said we were going to deliver'?

Those values can deliver. Dalgas shared with me an example of a healthcare provider that had an outage that created £12 million in costs. Switching to paper-based manual processes created delays and increased overtime spending on top of the cost of rectifying the technology outage. Paper and human workloads can be more costly than the cloud, and cloud workloads have to be analysed in light of alternative costs. 

Future workloads

In March 2023, when we wrote about rising CIO interest in Cloud FinOps to manage cloud costs, we were not to know that ChatGPT would lead to board-level demand for AI adoption. This demand will need to be factored into the cloud workload assessments of CIOs. As Miller says: 

The hype that there has been about Chat GPT has driven a lot of focus at the board level.

And CIOs cannot be on the back foot in terms of adopting this technology, Miller adds: 

We have a group established to coordinate AI across the business, and we have already found some interesting use cases for process efficiency. One internal process is a 90% performance improvement.

Petty at construction firm Laing O'Rourke is another that sees AI opportunities for augmenting processes within the international firm:

We are looking at text summarization tools for bid writers so that it relieves the pressure on teams to meet deadlines, and we have a proof of concept out at the moment.

Asked about meeting the AI demand from business lines, Petty says: 

Revenue generation is the first priority in identifying the opportunities, as we cannot attack all of them.

It is inevitable that AI, especially if organizations choose to develop their own LLMs, will increase cloud workloads and, therefore, costs. And technology leaders are already factoring this into their analysis. Miller of the insurance sector says: 

The cost of delivering AI-based services will become prohibitive, especially if the availability of GPUs at hyperscalers becomes a challenge. That may become a constraining factor.

Miller adds that the adoption of AI must follow the same path as DevOps and fail fast so that costs do not impact the bottom line adversely. 

Alongside the demand for AI, organizations continue to focus on data strategies, with the bulk of these using cloud-based data lakes and data warehouses. Again, this adds to the cloud workload management demands on CIOs. Petty says Laing O'Rourke has just completed a three-year Azure-based data warehouse and data platform program. 

My take 

Workload management is an indicator that the role of the CIO has changed in the cloud era. The CIO has to be an orchestrator, meeting the demands of a business that needs the cloud to digitize its processes and meet the needs of consumers or use the cloud for AI. 

Workloads operating in the cloud are clearly providing organizations with increased agility. Managing and optimising cloud workloads is increasing the position of CIOs as cost-saving and sustainability leaders - in other words, business and change leaders.