By Karan Kirpalani, Cloud Head, NTT Ltd.
The cloud has undoubtedly become a big force and is increasingly being relied upon for every digital strategy. Not surprisingly, Gartner says that end-user spending on public cloud services in India is forecast to total $7.3 billion in 2022, an increase of 29.6% from 2021. This is expected to accelerate faster in the next few years, with Gartner predicting that 85% of organizations will embrace a cloud-first principle by 2025 and will not be able to execute their digital initiatives without using cloud-native technology.
While enterprises benefit from the scalability and resiliency of the cloud, for many enterprises, keeping cloud costs under control is a challenging task. Gartner, for example, predicts that through 2024, 60% of infrastructure leaders will face public cloud cost overruns. Given the fact that the cloud will represent a growing share of the IT budget, and will soon eclipse the on-premises budgets, it is imperative that organizations adopt a proper strategy and monitor cloud costs using adequate controls.
We recommend some of the following steps:
Choosing the right cloud strategy
Depending on the application portfolio landscape, it is critical to clearly define the pre-migration and post-migration stages, and document the IT inventory (apps, databases, networks, interdependencies between apps, etc), and the expected outcome. Without clear objectives and understanding, it can result in higher operating costs after migration. For example, in the case of some applications, a lift and shift approach may be the most suitable, while for some applications, a re-factoring approach may be the best approach.
Understand application dependencies by using APM tools
Research firm, Gartner, for instance, predicts that by 2020, 75% of organizations will have deployed a multi-cloud or hybrid cloud model. That said, the hybrid cloud poses a unique set of challenges for enterprises. A hybrid IT infrastructure consists of multiple and sometimes disconnected IT applications which can cause service disruptions. As multi-cloud and hybrid cloud usage increases, many organizations are finding it extremely challenging to ensure performance and availability due to a lack of visibility into the end-user experience for both cloud and on-premises applications.
When infrastructure is deployed across multiple clouds, organizations find it challenging to take a comprehensive and unified view of application data. Failing to understand application dependencies, can lead to migrations taking more time than originally envisaged, leading to increased costs.
This is where an Application Performance Monitoring (APM) Solution can help and give enterprises the required visibility into every business transaction for a given application that may cross multiple types of cloud environments. This could include analyzing application performance issues that can be caused by the application itself or any associated elements such as the underlying infrastructure, middleware, databases, or the network.
By using APM tools, organizations can get full visibility into the performance of every application and dependency. They can also gain the ability to visualize end-to-end cloud application performance in real time and see for themselves how every application component, line of code, and important infrastructure resource is performing. Before undertaking a cloud migration, organizations can use APM tools to ensure that every application dependency and performance requirement is understood clearly.
Right-sizing architecture for the cloud
Most enterprises make the mistake of overestimating the capacity required on the cloud. For example, one common mistake is replicating the same on-premises infrastructure on the cloud, which can result in additional costs. Rather than just lifting and shifting applications on the cloud, refactoring or rewriting applications to help them run better in the cloud can help in saving money over the long term. Enterprises can also use APM tools to automate resource allocation based on performance for an accurate, consistent, and cost-effective approach to dynamically scaling instances. An application-aware provisioning capability can help in adequately provisioning throughout the technology stack, whether on-premise, hybrid, or a cloud-native architecture.
Monitor idle resources
Cloud waste is common, and many enterprises have virtual machines sitting idle. By using cloud monitoring tools, enterprises can turn off their virtual machines when not in use. For example, instances purchased on demand (used for non-production-related activities such as development, QA, and testing) are often left unused on nights and weekends. By turning off these instances during non-usage, enterprises can significantly reduce their cloud-related costs. This can be done using automated policies that shut down resources during pre-determined timings. Similarly, cloud monitoring tools can also use predictive analytics to forecast future capacity.
Besides the steps recommended above, organizations must also make individual business units accountable for the resources they use. This can lead to proper usage, as application owners will learn to be more aware of their infrastructure usage. To help in monitoring and drilling down to finer details, enterprises must also tag all cloud computing resources. Enterprises can also use cloud monitoring dashboards for monitoring and alerting on consumption patterns in various divisions. A managed cloud services provider can also be of great help. Most managed cloud services and solutions providers have the experience and expertise besides the necessary tools and technologies to help in keeping cloud costs under control.
Cloud computing is a game-changing technology if used well. However, it can also be a costly exercise, if organizations do not have the visibility and control over cloud costs. By following the steps outlined above, we believe your enterprise can ensure that cloud costs never spiral out of control.