Cost Takeout the Clever Way!

Does your IT budget leave no room to maneuver?

To use cloud computing in a compliant way and to keep control there is more needed than in the early days many years back. Can you remember the times when the primary method to procure cloud services was the personal credit card and the expense statement each month?

Nowadays, cloud computing is an established delivery model and there is a mix of company-wide frame contracts with central billing and more adventurous ways of procuring cloud services. One thing all of them have in common though – to use them, some pre-requisites need to be in place.

To get an idea about these pre-requisites in more detail, I do like to point you at the onboarding blog series where we covered these from a content perspective.

How to onboard a cloud solution provider – part one

The foundations of hooking up to a cloud provider – network, identity and integration

Making Cloud Onboarding hassle-free and valuable

Tips for Seamless Cloud Onboarding

The key question is how to cover all these changes from a resource and cost perspective. Your budget is already fully planned for, your team is overloaded anyhow, and the business is demanding new (cloud) services to be made available in record time. At the same time, the legacy environments take up more and more of your team’s time and budget and the IT strategy demands to move these to a ready-made cloud bed.

So, where can you free up both the resource and budget to prepare the bed? You have the lingering feeling that the cloud use to date has not driven visible cost savings but can neither pinpoint whether this is a fact, nor can you devise the reasons for this. You have visited the exhibitions at the events of the big hyperscalers and seen myriads of tools that promise cost takeout.

The crux with all these tools is that you need to buy a license or procure a services fee and that you need experts to implement these and make sense of the results. There is no one in your team you can free up to learn and bringing in outside help is another cost position your budget is not made for.

Before we solve this conundrum let’s look at some cost drivers that will be taken into account and the potential results.

The hidden cost optimization potential

The base premise of cloud, at least for IaaS, is that the unit cost is cheaper than of the self-owned, self-operated, maybe somewhat outdated server in your own building or even a co-location facility. So, where does additional cost slip into this cloudy pink vision?

1. Not using the best-suited service type from a cost perspective
2. Non-intentional misuse leading to extra cost
3. Relying on outdated sizing assumptions
4. Intentional use of high cost services

The service types of hyper-scalers do give you a range of options to manage the cost/performance balance. Yes, on the technical side scaling up and down is a huge benefit that can also have a positive impact, but the focus here is on the commercial variants. Whether that is the type of e.g., virtual machines and the question whether one instance with a larger processor is cheaper than two smaller ones or the so-called reserved instances. With the latter, a usage promise will lead to lower unit prices and this is one of the core areas for optimization, especially with stable workloads or at least the stable need of workloads that would use classic cloud cost models to ride the scale wave, e.g., at Christmas for retailers.

The next area is one that has to do with change that goes beyond technology. Let’s look at a developer in the pre-cloud days. The development environment was on a server shared by several developers. It was being used and since it was paid for already (let’s ignore energy), it ran day and night. No extra cost was racking up even though from a unit cost perspective (HW investment, operational cost, maintenance, etc.) this model was horrible. Now we introduce cloud into this scenario. The unit cost goes down and each developer gets his own environment. Other than with capex driven model in the cloud world, each minute of use costs money. So, it makes a huge difference if the environment is being shut down at the end of a shift. Any testing should happen in the dedicated test environments and there is no need to keep a development environment running 24*7. This is an area of huge cost saving potential and there are two ways to drive the change. Education and holding the developers accountable and automating a shutdown at defined shift-end events. It is important for the latter approach that there is a warning prior to shutdown. Also automated backups are critical. It is a misuse of cloud computing that is not intended to do harm by the developer but rather sticking to the old ways and keep stuff running.

Another area is the direct result of the easiness of deploying services in the cloud. You will see an increase in deployments and that is part of the cloud value. From a cost perspective though, you must tightly monitor usage. Consumption of a service can go down significantly after a short period when the trial phase is over, the “new” factor wears off, or because functionality is being offered in other areas. Shutting down low consumption services, temporarily if the consumption is seasonal, or permanently if it is just not, will free up the budget in a significant way.

In addition, licensing can become a factor in an environment where you deploy more instances. Closely watch the licensing rules and models and consider moving to open source models, e.g., from Oracle to Postgres.

We have established by the sample areas which are just a small sample, that there is potential for optimization. Let’s now investigate that.

Optimizing Cloud Cost without upfront investments

If your team is busy and your budget is tight you might acknowledge the potential, but your hands are bound to tap into it. You know that the gradual way into the cloud is one of the reasons that left you lacking an integrated cloud management but you cannot correct this. Not only missing the cost saving expectations but actually having higher costs is increasing the pressure on the team and you personally.

In short, you are backed into a corner.

Getting out of the corner does mean that you realize the cost optimization potential, which would create a budget to act and fulfill the organization’s expectations. To do so you need to realize it without an upfront investment. No investment into licenses and experts. But this is not just about a deferred payment which you would need to budget for anyhow and not help you. The question is how can you pay for the effort from the outcome? That reduces the risk that you spend money and do not realize the promised potential and it ensures that you can pay it out of the future budget that already reflects the savings.

Do these models exist? Who would commit to you to follow such an approach? Yes, the models do exist.

Outcome-based optimization projects that cover three major areas are available in the market.

• Baselining the existing usage
• Discovering the optimization potential
• Recommending changes to realize the potential

Outcome-based means that the partner is committing to a percentage of the saving. You and your partner jointly define the scope and the time frames and only after the potential has been realized successfully, they get paid. To apply such a business model the partner must be confident about the optimization approach and totally committed to your success.

And if the partner is good, the consultants also point out that such a cost optimization exercise r is neither a one-time activity nor should you do such projects on a regular basis. The best way is to clean up and then implement a cloud management layer that drives cost optimization continuously. The best model will be a model that uses AI to constantly evolve and a partner that not only brings a tool but also configurations and accelerators to make the most out of it.

The cultural change and the impact on the strategy on optimized use of cloud and an optimized budget have is the next step in the journey. First you create wiggle room and then you get better from thereon.