What is cloud cost optimization?
Let’s start with two scenarios.
Scenario 1
Imagine you are a retail bank. One of your credit applications has several modules frequently used by in-house teams and your banking customers. Out of all those modules, you have recently migrated some of the modules majorly used by the back-office and analytics teams to the cloud while the customer data and some critical parts of those modules are still sitting in your on-premises server.
Right now, your estimated cost for those applications is standing at $1 million. The cost covers compute, storage, network, managed DB, and other additional services.
After two months, you realise that not all the compute instances or storage capacities are fully utilised. You discover several idle elastic IP addresses. On the other hand, some critical resources are running all the time, unnecessarily. These unused, underused or overused capacities are rapidly drying up your capital reserve. You must find a way to provision your cloud resources to achieve cloud cost efficiency.
Scenario 2
Now imagine you are a global leader in providing software for workplace management. Due to your company’s aggressive expansion plans, it has acquired many firms over the last few years. For your IT team, this global acquisition endeavour has translated into managing 70 different cloud accounts across all leading hyperscalers. Whereas for your organisation, budgeting and ownership have gone for a toss, directly affecting the decision-making.
You have to chase more than 25 people every time to get reliable figures. Not just the numbers, the way they perceive cloud usage also varies greatly. You must find a cloud provider solution that would offer you unified visibility of all your multiple clouds accounts and right practices for cloud cost management.
Looking at both the scenarios above, we can understand that switching to a new cloud environment is just one tiny part of the story; how you manage your cloud spend and reduce your cloud waste are the bigger and more crucial aspects. And that’s where cloud cost optimization comes into the picture.
Cloud cost optimization
Cloud cost optimization is a process that combines tools, techniques, best practices, and strategic roadmaps to optimize cloud costs of an organization while maximizing the value of cloud services.
At its core, cloud cost optimization strategies guide IT teams and decision-makers in running applications in the cloud effectively and efficiently at the lowest possible cost.
“80% of enterprises consider managing cloud spend a challenge.”
~ McKinsey & Company, 2020
How to optimize cloud costs
There are two distinct areas that an organization must master in order to bring cloud optimization and efficiency.
2 key areas of cloud cost optimization
Procurement
Can be achieved by leveraging tools and low hanging opportunities to save cloud bill
Optimization
Can be achieved by optimal configuration of cloud resources and workloads
As a practice, cloud cost optimization identifies mis-provisioned resources, cancels unnecessary resources, takes advantages of any saving opportunity, finds ways to reserve higher capacity resources at lower costs, and rightsizes cloud resources to match certain workloads and applications.
Why is cloud cost management so important?
There are specific business goals that drive every cloud migration. The cost of managing on-premises infrastructure remains a key one. Hence, every move to new cloud environments is aimed at optimize spending or minimize the cost overhead i.e. the cost of producing products or services.
Take the first scenario for example. Every component of the compute, storage, network, and managed DB must be provisioned in a way that the credit application can support the massive scaling needs of a super sale day or festive times.
At the same time, each component of the application modules must have a number of servers, database clusters, and other instances running across geographically spread data centers. Some of the instances can have active-active configuration, or active-standby. Without any cloud cost optimization strategy in place, the credit company will be paying for unused resources or instances which are sitting idle most of the time. And we are talking about billions of dollars of resource usage here.
The list does not end here. For on-premises infrastructure, the challenges are different. Often one application shares infrastructure, firewalls and more with other applications. This shared infrastructure is provisioned to handle the biggest scale-ups. To support that system and the anticipated surge in workload, IT teams invest in purchasing storage and backup, physical space, power and cooling system, and more. Now, if you just lift and shift this entire infrastructure and its components without any system redesign, you are bound to go beyond your cloud spending budget in the long run.
As much as 70% of cloud expenses are wasted.
~ Gartner
You might as well point towards one of the biggest selling points of cloud infrastructure—the pay-as-you-go models. But, pay-as-you-go can be a tricky affair unless you have experts guiding you throughout the changes, potential cost savings, nuanced licenses of different cloud services, cloud storage, cost centers, or cloud platforms, how to avoid vendor lock in, etc.
That is the technological aspects of cloud costs. Now comes the operations aspect of cloud bill. In every organization, primarily one team is responsible for allocating cloud resources—DevOps or CloudOps.
Most of the time, DevOps people are caught in a tug of war between finance teams and application users or owners. The finance wants cost controls and save money at any cost, and the application owners are not ready to let go of idle resources. How do you find a middle ground by meeting the needs of both these stakeholders?
The answers to these above scenarios and questions lie in cloud cost optimization. By leveraging cloud optimization techniques, DevOps can easily find ways to maximize the value of cloud against spend, enough to make the CFO and finance happy, and ensure delivery of seamless performance at enterprise scale, making the application teams as well as the end users satisfied.
Add to that the opportunities cost optimization presents ranging from a single window to manage and monitor all cost heads or multi cloud cost management, gaining cost efficiency through power multiplication, reducing application risks by rightsizing instances to timely workload routing and reservation, optimizing software licenses, and more.
4 steps to optimize costs of your cloud environment
Create a framework
Improve spend visibility
Reduce wastage
Optimize resources
Create a plan
Increase spend visibility
Optimize resources
Decrease spending
8 best practices to optimize costs
When structured well and implemented strategically, leveraging the best cloud cost optimization tools and techniques, it can control cloud costs and truly save significant amounts of money for organizations. To help enterprises understand the complex costing and provisioning game, all the hyperscalers and cloud providers have built frameworks and the most effective practices, e.g. AWS or Azure cost management best practices. However, in multi cloud cost management scenarios, these best practices get overlapped, making the process even more complex.
How to reduce cloud bills is a learning curve that experts master over a period of time with trial and error. Fortunately, there are several best practices with proven results that continue to help IT leaders in the right direction to reduce costs or create cost awareness. Here are 8 cost optimization techniques from the experts:
1. Set up a master account
Decentralized decision-making is a key reason behind soaring cloud billing. The first and foremost step would be to solve this challenge by setting up a master payer account or central cost center. This master account or central cost center brings higher cloud cost visibility across cloud providers as every time you add new members to the account, all the cost information reflects in the master account. It acts as a single source of truth, offering you a singular picture of your organization’s total cloud spend. At the same time, it enables leaders to track and reduce costs in the long run.
The next step is to understand what’s happening in your cloud system. To capture context, use readily available telemetry tools such as Amazon Cloudwatch, Microsoft Cloud Monitoring, Azure Application Insights, AppDynamics, Redgate, CloudTrail, etc. It’s critical to understand this context and map it against the cloud pricing. The third and last step is to keep a tab on your cost history, review past billings, and identify cost anomalies.
2. Ensure your internal budgeting is in line with business goals
Another major challenge arises when everyone is not aware of their budgets and business goals for each of the projects. Engineers and developers end up choosing arbitrary figures. This challenge can be solved by ensuring the leaders directly communicate with the appropriate teams, such as the product leadership team on budgets as well as cost requirements.
Some of the questions that a cost requirement should answer are:
- Packaging and delivery of product features
- Features to be included in paid subscription or free trial?
- Speed and resiliency of delivery
3. Make cloud cost a key metric
Making sure everyone involved in cloud computing in your organization understands what goes into the billing can be a stepping stone towards optimizing costs. Here’s how you can make cloud costs less intimidating and easier to understand:
- Take current numbers and scenarios
- Explain the context
- Explain cloud providers offer discounts
- Demonstrate various storage tiers from a cost angle
- Show how to each storage tier impacts the billing
- Show how to measure each of the cost components
- Define good cost and bad cost
Key metrics pertaining to the cost:
-
Unit Cost
- - Cost per API call
- - Cost per report
- - What factors affect the unit cost
- - Impact of this cost on the bottom line
-
Idle cost
- - Zero customer load
- - Can measure efficiency
- - Helps determine architectural changes
-
Shared infra
- - Offers cost savings
- - Enables engineering efficiency
- - Splitting costs is a challenge
- - Roadmap for chargeback is necessary
-
Cost/Load efficiency curve
- - Understand growth of cost/load curve
- - Important to recognize any sudden growth in advance
-
Innovation to cost ratio
- - Determine R&D cost ratio to production operation costs
- - Otherwise hinders the process of moving to production
-
Unit Cost
- - Cost per API call
- - Cost per report
- - What factors affect the unit cost
- - Impact of this cost on the bottom line
-
Idle cost
- - Zero customer load
- - Can measure efficiency
- - Helps determine architectural changes
-
Shared infra
- - Offers cost savings
- - Enables engineering efficiency
- - Splitting costs is a challenge
- - Roadmap for chargeback is necessary
-
Cost/Load efficiency curve
- - Understand growth of cost/load curve
- - Important to recognize any sudden growth in advance
-
Innovation to cost ratio
- - Determine R&D cost ratio to production operation costs
- - Otherwise hinders the process of moving to production
4. Define other key metrics
Similar to what we just discussed in Best Practice 3, it is crucial that whoever is impacting the cost has a clear understanding of the business and its goals in order to ensure better decision-making. At any given point in time, the most critical business goals should be clear for all stakeholders. It can be an increase in margins for large enterprises and customer growth for startups. If there is one team that should always have complete clarity of the goals is DevOps as they are the ones involved in making day-to-day decisions for both the business and its customers. Here are 11 key business metrics to start with:
- Cost per feature
- Cost per customer, segment
- Cost per app/platform
- Cost per team
- Cost per unit
- Cost per cloud service
- Revenue
- Cloud cost
- Cost of R&D
- Time to market
- Cost deviations
5. Ensure each team has access to relevant, useful data
Too much irrelevant data creates noise that can do little good for the organisation. Hence, one must ensure that the relevant data reaches the relevant team or people at the right time. Take the engineering team, for example. This team needs data to compare. For DevOps teams, the value lies in baseline data that helps them identify anomalies, analyse previous billings, track if something is broken, etc. They also need data to slide and dice by other teams, by individual resources, by service, or by feature. With better access to such granular data, DevOps team is better equipped to solve problems faster.
On the other hand, it’s the job of the finance to look forward in order to plan well in advance. They want to know the cost predictions, the future estimations if the number of customers goes up, or the anticipated rise in the expenses in the coming fiscal year. They also care highly about the ROI.
This means that both DevOps and finance teams are viewing the same data. But they need it in different formats in order to derive insights from it. Insights that are relevant to their roles and will help them perform better.
6. Make data your best friend in optimizing cost efficiently
How often have you given cost a thought after a product/feature has been ready for launch or already launched? This is a common mistake many large organizations make without understanding different pricing models or storage tiers or their financial implications.
Make cost saving a key consideration throughout the product development lifecycle if you don’t want a shock at the end of the cycle. Here’s how it can be done:
Planning
The budget for the product should be justified by the teams. The product roadmap should also include the technical-debt data. This way, any unexpected spend can be reduced and if necessary, the budget can be revised.
Deployment and operation
A pre-defined cost roadmap helps teams in quick identification of any unplanned spend and adjust the budget accordingly.
Design and Build
A clear understanding of the costs helps teams make cost effective decisions regarding architecture in this stage. It’s important they have the right cost data in hand to understand the unit costs and more.
Monitoring
In the last and final stage of the product lifecycle comes the importance of reassessing costs by product/feature/team and create an extensive report on OpEx and ROI.
7. Capture the complete picture in one frame
The lack of visibility into cloud spending plays a major role in skyrocketing the unexpected costs for multiple teams working with multiple cloud computing dashboards and platforms. It not only impedes the decision-making process but also affects the provisioning and rightsizing of cloud resource. A single source of truth can solve this challenge by bringing all the petabytes of data from different cloud providers together and presenting it in an insightful manner.
This unified window of cloud costs offers teams the much-needed visibility into the cloud spend, multiple accounts, and resources. Moreover, it allows them to zoom in on particular accounts or resources to get granular insights. As a result, every cost spent per customer, per feature, per cloud resource, and per deployment becomes visible to the leaders.
8. Make cost optimization a continuous process
Cost optimization is not a one-time process. It has to be imbibed into the company culture and established as a continuous process. Along with it, organizations must invest in creating and nurturing a cost awareness culture top-down. How to make cost optimization a continuous process?
- Standardize cost optimization best practices for cloud operations
- Assign teams or specific resources with cost governance responsibilities
- Encourage accountability and continuous improvement across the organization
- Make maintaining an efficient cloud system a top priority for all
Final thoughts: This is the time to pave way for cloud cost intelligence
All the above practices need a massive amount of data from different teams, accounts, projects, features, and resources. While there are thousands of tools to bring down your cloud costs, only a few can provide you with the exhaustive data required to effectively and efficiently implement and monitor these best practices.
As an IT leader, you are required to find solutions that not just look backwards but also look forward and guide you in aligning your costs to the business goals. Therefore, today, the goal is to achieve cloud cost intelligence through cloud cost optimisation to stay one step ahead.
