AWS Partner Network (APN) Blog
How VGS achieved cost savings on Amazon Aurora
By Pavlo Ulanov, Senior Infrastructure Engineer – VGS
By Bo Niu, Senior Solutions Architect – AWS
By Abul Khan, Technical Account Manager – AWS
 |
| VGS |
 |
Organizations are looking to transform and optimize their database usage. VGS harnesses Amazon Aurora as their primary relational database service, powering their cloud-native infrastructure. After conducting a thorough cost analysis of their Aurora implementation, VGS discovered an opportunity to optimize their database spending.
In 2023, Amazon expanded its managed relational database service, Amazon Aurora (MySQL and PostgreSQL compatible), with the introduction of Amazon Aurora I/O-Optimized. For applications needing high I/O throughput, the updated I/O-Optimized storage provides enhanced performance, decreased latency, and consistent pricing. Common use cases include e-commerce and payment processing systems, which align well with VGS’s needs.
The introduction of Amazon Aurora I/O-Optimized by AWS, promising better price/performance ratios, caught VGS’s attention as a potential cost-optimization opportunity. In this blog, we’ll show you the steps VGS took to reduced their database spend on Amazon Aurora by moving to Amazon Aurora I/O Optimized.
About VGS
Founded in 2015, VGS is the world’s leader in payment tokenization and trusted credential management platform, depended on by Fortune 500 companies, merchants, fintechs, and banks alike. VGS’s mission is to revolutionize the way sensitive data is stored and secured, enabling organizations to manage information across cards, bank accounts, and digital wallets with ease.
VGS stores sensitive data and tackles critical payment acceptance challenges such as multi-PSP management, card issuance, payment orchestration enablement, PCI compliance, and the protection of personally identifiable information (PII). We provide our clients with complete ownership, control, and insights into their payment data, driving growth and enhancing user experiences across industries.
VGS offers a comprehensive suite of solutions, including a composable Card Management Platform, a PCI-compliant Vault, and advanced network value-added services such as Network Tokens, Account Updater, and Card Attributes. Our innovative technologies empower businesses to boost revenue through higher authorization rates, reduce fraud, and streamline operations—all while seamlessly integrating with existing tech stacks.
Identifying and Optimizing the Most Expensive Amazon Aurora Cluster
VGS team analyzed VGS’s Amazon Aurora spend using Cost Explorer and found that Aurora:StorageIOUsage accounted for 30%-40% of VGS’s daily costs. Amazon CloudWatch metrics (VolumeReadIOPs and VolumeWriteIOPs) helped us identify the most expensive clusters. VGS team observed that only one or two clusters had millions of I/O operations, with a daily spike on Mondays due to a cleanup cron job as shown in Figure 1.
Figure 1 – I/O DataFileRead spike on Monday
Performance Insights shown in the above graph revealed that IO:DataFileRead events matched the CloudWatch VolumeWriteIOPs and VolumeReadIOPs metrics in Figure 2 below, confirming the spike pattern.
Figure 2 – Sum of VolumeWriteIOPS and VolumeReadIOPs spike on Monday
VGS team identified the most expensive Amazon Aurora cluster, aligning CloudWatch and Performance Insights data, and plan to migrate it to Aurora I/O-Optimized first.
Transitioning to I/O Optimized Storage Configuration
Utilizing infrastructure as code (IaC) streamlined VGS’s transition to the new storage type. The VGS team executed the change during the Amazon Aurora maintenance window, ensuring minimal disruption. The console switch to modify the existing Amazon Aurora cluster was seamless, requiring just a single command through IaC and no downtime. The VGS team encountered no issues during or after the switch. The impact was immediately visible in Cost Explorer. In the hourly view, the daily/weekly cost spikes vanished as shown in Figure 3 below.
Figure 3 – Cost Explorer showing cost spikes was eliminated post deployment after Sep-22
The Aurora:StorageIOUsage can be seen in blue color in above graph. You will notice sudden drop after Sep 22. Additionally, the weekly usage spikes were eliminated. Also, the overall cost is reduced. You might think, ‘why there is still IO charge past Sep 22, although very low percentage?’ This is because VGS still had some other clusters in standard configuration. You want to pick IO optimized for specific clusters where I/O spending accounts for 25% or more of total Aurora database spending.
When contemplating the switch to I/O-Optimized storage mode, it is crucial to take several factors into account.
- Ensure that your cluster operates on a compatible engine version and instance type.
- Changing storage types is limited to once every 30 days.
- A thorough evaluation is necessary to determine if I/O-optimized storage is the right fit for your needs. AWS advises
- considering this option if your I/O expenses surpass 25 percent of your current Amazon Aurora database costs.
- Additionally, staying updated with the latest official AWS documentation is essential to make an informed decision.
Achieved substantial cost savings
Transitioning to I/O-Optimized storage mode for Amazon Aurora has proven to be a game-changer for VGS, leading to substantial cost savings in the hundreds of thousands of dollars without any negative repercussions. The overall expenditure on Amazon Aurora saw a notable reduction of approximately 20%.
Visit Aurora storage and reliability to learn more about Amazon Aurora IO/Optimized. Check out VGS on the AWS Marketplace.
.
.
VGS – AWS Partner Spotlight
Very Good Security (VGS) is an AWS Financial Services Competency, AWS PrivateLink Service Ready and ISV Accelerate SaaS Marketplace partner that protects sensitive financial information, orchestrates payments, and consolidates transaction data from multiple processors.