Hello, cloud enthusiasts! As a Technical Account Manager at AWS, I've observed that many customers miss out on unique cost-saving opportunities. Here are some unconventional tips that can lead to significant reductions in your AWS expenses
Spot Instances for Batch Processing
Let’s me share an experience of working with leading financial services client. They had a rigorous data processing pipeline that ran nightly jobs to analyze market trends. Initially, they were using on-demand EC2 instances for these tasks, and the costs quickly added up—over $15,000 a month!
During a cost optimization review, I suggested they explore Spot Instances. After explaining how Spot Instances can significantly reduce costs for flexible workloads, we ran a test. They migrated their nightly batch jobs to Spot Instances, leveraging their existing job scheduling framework.
The result? They saved around 80% on their compute costs, bringing their monthly bill down to approximately $3,000! The jobs ran seamlessly, and they were able to take advantage of unused EC2 capacity without compromising on performance.
Tip: Assess your workloads for flexibility. If you can tolerate interruptions, consider using Spot Instances for non-critical batch processing jobs. This simple switch can lead to significant savings.
Optimizing UAT Costs
Recently, I worked with a customer who had just migrated a major application to the cloud. During our monthly cost analysis, we found that their UAT costs consistently exceeded their production costs, even after going live. When we inquired, the customer mentioned that developers were conducting tests on multiple instances that were running 24/7. Additionally, they were maintaining a multi-AZ setup (exact replica of prod environment) for all their RDS and EC2 instances in UAT.
Recognizing a significant opportunity for savings, we collaborated with the customer to implement automated scripts that scheduled the start and stop of instances based on tagging. This adjustment alone reduced their compute costs by around 50%! Furthermore, we recommended switching all RDS instances to single-AZ configurations and adjusting their application to utilize only two availability zones instead of three, which further contributed to cost savings.
Tip: Regularly analyze your UAT environment for opportunities to optimize resource usage. Implementing automation for instance management and evaluating your architecture can yield substantial savings.
Migrating EBS Volumes to GP3
One particularly enlightening scenario involved a customer running multiple io1 and io2 type EBS volumes. During our analysis, I asked how much IOPS they really needed, and they estimated around 2,000 IOPS per volume. This led us to discuss the benefits of GP3 volumes, which offer up to 3,000 IOPS at a significantly lower cost.
By migrating from io1/io2 to GP3, the customer was able to save about 50% on their EBS volume costs while still meeting their performance needs. It was a win-win situation!
Tip: Regularly evaluate your EBS volume types and consider migrating to GP3 if your IOPS requirements allow it. This simple change can lead to substantial savings.
CloudWatch Log Management
One of the more common pitfalls I’ve seen is how clients handle logging in CloudWatch. A customer I worked with had enabled detailed logging across their UAT and production environments—everything from Lambda invocations, ec2 instances , EKS control plane logs to API Gateway requests. They didn’t realize the volume of logs would balloon their CloudWatch costs to nearly $30,000 a month!
After reviewing their logging strategy, we helped them disable unnecessary logs in UAT and fine-tune their logging level in production, retaining only essential logs. The result? They managed to cut their CloudWatch costs by over 70%!
Tip: Regularly review your CloudWatch logging settings. Enable only the logs you truly need and consider adjusting log retention policies to manage costs effectively.
Monitoring Reserved Instance Utilization
While conducting a monthly cost analysis for a leading eCommerce customer, I discovered that their DB Reserved Instances (RIs) were not being utilized. When we reached out for clarification, they were puzzled since they were already running the instance. Upon further investigation, we identified that they had accidentally selected the wrong OS type during the RI purchase, preventing the RI from being applied.
This experience highlighted the critical importance of monitoring RI utilization through budget alerts and double-checking the parameters at the time of purchase, especially since RIs are non-cancellable.
Tip: Always verify the parameters when purchasing RIs and set up alerts to monitor their utilization. This diligence can prevent unnecessary expenses and ensure you're getting the value you paid for.
Modernizing Instances to the Latest Generation
Many customers often overlook the fact that new generation instances are typically more cost-effective than older ones. During our cost optimization discussions, I emphasize the advantages of transitioning to the latest generation instance types, which not only deliver enhanced performance but also come with better pricing.
By modernizing to the latest generation, one customer saw a notable increase in performance along with a reduction in costs.
Tip: Regularly assess your instance types. Moving to the latest generation can provide both performance enhancements and significant cost savings.
You might think collecting all this data seems overwhelming, but you can simplify the process with AWS Cost Optimization Hub Dashboard. The dashboard experience gives you a quick overview of all the recommendations with estimated savings without duplication. You can zoom into any opportunities by filtering and grouping with fields such as account and region, and prioritize them by savings and implementation effort.
If you are running your workloads on AWS and need any support, leave a comment below. I would be more than happy to assist you in optimizing your AWS costs
Top comments (0)