Rightsizing for cloud cost optimization can be one of the most effective ways to reduce cloud costs. Most companies understand how important it is to right size resources in the cloud to optimize cloud infrastructure and achieve cost-efficiency.
However, it can be difficult for the company to identify which resources are running, at what capacity they are running, and by whom, to effectively right size their environment. This becomes more challenging as companies increase cloud usage or implement a multi-cloud strategy.
In this article, we explain the meaning of rightsizing in the cloud, how to identify opportunities for rightsizing, how it helps to reduce cloud expenses, who should use rightsizing, and how to implement an effective rightsizing strategy to reduce cloud bills and gain operational efficiency in the cloud.
What does rightsizing mean?
- Analyze the utilization and performance metrics of your infrastructure, (volumes and virtual machines)
- Determine whether these are running efficiently, and what actions you could take to improve efficiency
- Modify the infrastructure as needed (ie. upgrading, downgrading, terminating)
- Upgrading is recommended for workloads with high utilization. The performance metrics should be analyzed over a significant time-period to reach this conclusion.
- Downgrading or downsizing, is recommended for resources that are underutilized and achieve the same performance, even with a downsized workload.
- Terminating is recommended for assets that are running within your account but are not in use, also known as “zombie” assets. Terminating these assets tend to result in immediate cost savings.
A simple example is if your cloud environment is oversized – you have 50 Virtual Machines (VMs), but each individual VM is not being completely utilized, therefore the workloads that are involved could run effectively on fewer. In this example, the right approach may be to simply delete or resize some instances.
Free cloud cost optimization. Lifetime
How does rightsizing help reduce cloud expenses?
Rightsizing is the most effective way to control cloud costs. It involves continually analyzing instance performance and usage patterns, and then terminating idle instances and rightsizing instances that are overprovisioned to the workload. Since companies’ resource needs are continuously changing, rightsizing must be an ongoing process to achieve cost optimization. Users can make rightsizing a smooth process by establishing a schedule for each team, enforcing tagging for all instances, and taking full advantage of the powerful tools that clouds like AWS, GCP, Microsoft Azure and others providers.
Below we discuss various approaches, and benefits of each, to effectively rightsize instances:
1. Rightsize based on usage patterns
As companies monitor performance, they can identify the usage patterns to maximize potential rightsizing options. For example:
- Steady state – when the load remains at a constant level over time, and the compute load can be accurately forecasted
- Variable yet predictable – The load varies, but on a predictable schedule
- Development, testing – These environments are typically used during business hours. They can be turned off when not needed such as during evenings or weekends.
- Temporary – Temporary workloads with flexible start times. Here users can consider placing a bid for a Spot Instance rather than an On-Demand Instance.
2. Turning off instances that are idle
One of the easiest ways to reduce operational costs is to turn off instances that are no longer in use. If instances are idle for two weeks or more for example, it should be safe to terminate them. But before terminating an instance that has been idle for a couple of weeks, a few things should be considered. For example, who owns the instance, what is the impact of terminating the instance, and how difficult is it to re-create the instance if it is later necessary to restore.
Another way to reduce costs is to stop instances used in development and testing during hours when these instances are not in use. When their capacity is again needed, they can be started up again. Assuming a 50-hour work week, companies save up to 70% by automatically stopping these instances during non-business hours.
3. Rightsize by selecting the right instance family
You can rightsize an instance by migrating to a different model within the same instance family, or by migrating to another instance family. When migrating within the same instance family, all that needs to be taken into consideration is vCPU, memory, network throughput, and ephemeral storage. If your maximum CPU and memory usage are less than 40% over a month for example, you can safely cut the machine in half, saving your money.
When migrating to a different instance family, it’s necessary to ensure that current and new instance types are compatible in terms of their virtualization type, network, and platform. When you resize an instance, the resized instance has the same number of instance store volumes as when you launched the original instance.
4. Rightsize your database instances
You can scale your database instances by adjusting memory as performance and capacity requirements change. When scaling a database instance, consider some of the following:
- When you scale your database instance up or down, your storage size remains the same.
- You can modify your cloud instance to increase storage space or improve performance by changing the storage type
- Before you scale, make sure you have the correct licensing in place for commercial engines, ie. SQL Server, Oracle, etc.
- Determine when you want to apply the change, and if you want to apply it immediately or during the instance’s maintenance window
Who should use rightsizing and how to use it successfully?
Our experience shows that companies that rightsize their workloads can cut costs by as much as 30% to 70% when migrating them to the cloud. Even companies without licenses that can be migrated to the cloud can generate considerable savings by rightsizing their workloads. Companies often find that they can better provision their workloads if they establish a reserve of excess capacity that can be utilized when there are surges in activity, also known as, ‘burstable instances’.
Companies that rightsize the right way carefully assess usage patterns. They study the duration of average peak computing demand, to make informed decisions about downsizing server capacity and accommodating demand. They look at storage patterns, evaluating average memory usage and their proximity to the server to enable them to downsize more effectively.
Smart companies also recognize that migration doesn’t handle all workloads in the same way. Companies are likely to save more when they migrate nonproduction workloads to the cloud than when migrating production workloads. This is because computing usage in nonproduction workloads tends to be more volatile, and therefore more likely to benefit from the flexibility of AWS, Microsoft Azure, Alibaba, and other cloud providers. Additionally, companies find that they achieve greater savings migrating non virtualized workloads as opposed to virtualized workloads. This is because the benefits of virtualization are like those offered by the public cloud, as they relate to efficiency and flexibility.
Case in point: rightsizing can make a dramatic difference
TSO Logic, a Migration Competency partner of AWS, studied this event through a statistical analysis and sampled 105,000 operating system instances across North America. The report (Rightsizing Infrastructure Can Cut Costs 36%, by Mark Schwartz), revealed that only 16% of the OS instances were appropriately provisioned for their workloads. The remaining 84%, ran on larger than required footprints, which would amount to huge losses if the infrastructure was replicated directly to the cloud. The research then analyzed the effect that rightsizing would have had on the instances, based on real-world usage scenarios.
They discovered that organizations would have been able to reduce AWS cloud spend by 36% through optimally sized resources. In other words, rightsizing alone would have saved them more than $55 million per year. This analysis clearly shows the power of rightsizing. It can save organizations significant amounts in wasted cloud spend by optimizing instances, thereby retaining only the resources that workloads require.
Overlooked resources are contributed to a company cloud bill, and users don’t even expect that they’re paying for them.
💡 Find the ways of identifying and cleaning up orphaned snapshots to keep MS Azure and Alibaba Cloud costs under control → https://hystax.com/finops-best-practices-how-to-find-and-cleanup-orphaned-and-unused-snapshots-in-microsoft-azure-and-alibaba-cloud