Cloud spending is on the rise. The cloud’s ability to scale resources up and down to align with current demand requirements, often referred to as elasticity, is simultaneously both one of the key benefits of cloud computing and its greatest liability.
One thing that stands out amongst the enterprises that have truly realized the potential of the cloud against the ones that have not, is cost management. Many of the enterprises are not able to manage cloud spend efficiently and end up losing the cloud benefits. Though a few enterprises are able to manage the cost, they face challenges in mapping their cost consumption by different teams, products, and services and lack accurate chargeback and budgeting mechanisms. The most efficient and data-driven enterprises are the ones that have found ways to use their cloud spend for effective allocation of budgets to suitable products and services from a business perspective. This is referred to as cloud unit cost analytics.
Cloud resources are mostly shared in enterprises. The degree to which sharing happens varies from enterprise to enterprise. Few enterprises segregate teams and products using separate accounts, VPCs, dedicated instances, and so on. Despite these, there are cases where resources are shared between teams due to cost efficiency and optimum utilization. With such shared architectures, it becomes more difficult to break down the costs across teams.
Unit Cost Analytics (also known as Unit Economics) refers to the ability to provide the cost of specific business metrics, like cost per user, cost per API, cost per search etc., mapped to underlying cloud costs. With access to this, an organization can empower their teams to be more successful, self-managed, and optimized in taking control of their cloud costs. Enterprises that have command over their unit costs have a better command over their gross margins; because these organizations know that a steady or decreasing unit metric trend, even while costs increase, is an indicator of healthy growth. Measuring unit costs in cloud environments accurately helps in predicting the impact on future cloud costs as per changes in business demand. This also helps businesses to allocate budgets to products and features in an informed manner. Let’s look at a few real-life scenarios:
- An online fraud detection product might measure their cost per analyzed financial transaction, which is not only tied to the company’s cost to serve but also to how they price and package their products.
- An online travel reservation company might measure the cost of creating a reservation to learn how their costs change and forecast their costs during more popular travel times.
- A Rideshare company might measure the cost per ride by the time of day to determine the most cost-effective times to transport customers.
There’s no one-size-fits-all approach to implementing cloud cost unit analytics. The implementation depends on the business metrics, cloud environments, cost management, tools and practices, and many other important factors.
Let’s look at some of the important determinants that enterprises should consider for evaluating cloud spend –
- Business KPIs – Depending on the critical capabilities of the product or system, business KPIs that impact revenue, user experience etc., should be captured. For example, for an e-commerce application these could be cost per visitor, cost per order processed, cost per item searched, and so on.
- Mapping user/transaction journey with underlying cloud services and resources – Teams need to establish a clear path of user actions across services like front-end, middleware, and back-end. These layers are mapped to cloud services and resources.
- Tag compliance – Tag hygiene is extremely important to accurately allocate costs to cloud services (both dedicated and shared) to business capability. Like, an online order might involve front-end load balancers, middleware compute, back-end databases, integration queries and so on. Proper tagging is required across each of these resources.
- Shared services and resources – Clear identification of shared resources across use cases so that costs can be accurately distributed based on resource utilization, load, application metrics, and other factors.
- Telemetry – Telemetry metrics are an essential means to correlate application or system metrics with the cost of cloud resources. For e.g. the number of order processed would be a function of the number of API calls, queries. These metrics can be measured using telemetry data and tools.
Implementing cloud unit cost economics involves seamless integration across multiple tools like telemetry, application performance management (APM), log analytics, cloud cost management, and budgeting. The majority of the time, enterprises develop custom solutions using existing tools to get these unit cost metrics. This helps to quantify the impact of cloud spend on business performance at granular levels. Additionally, it promotes data-driven decision-making for product strategy and pricing decisions. This can also be used for uniform broad measurements to compare the performance of products and teams and anticipate cost changes due to architectural changes as per the forecasted and expected demands and new capabilities.
Contributed By – Mr. Mustufa Batterywala, Director of Engineering – Solutions, Gathr Data Inc
