About the Client
A UK-based FinTech providing CRM services for mortgage, lending and banking providers encountered a rare database compatibility issue after migrating to Amazon Web Services (AWS). It needed to resolve this matter quickly to keep cloud cost management under control.
Our managed services team enabled the client to bring actual cloud spend back in line with what was originally forecast. Next, they took additional steps to bring further cost, performance and security benefits.
- Using Amazon RDS for MariaDB (instead of Amazon Aurora) improved query management and cut costs.
- A bespoke DevOps anti-pattern for EKS cluster monitoring reduced costs by 99% compared to AWS’ out-of-the-box offering.
- This intelligent approach to cloud operations enabled the business to dedicate more resources to product innovation and improvement.
Challenge: Cloud Costs Spiral Due to Database and Monitoring Problems
As a scale-up CRM provider for the financial services sector, this organisation was keen to leverage the resilience, security, and cost-efficiency of the cloud. However, its containerised migration to AWS using Amazon Elastic Kubernetes Service (EKS) didn’t initially deliver the target outcomes.
Amazon Aurora had been recommended as the most appropriate relational database service to meet the client’s needs, but it was incompatible with the application. This led to inefficiencies in queries, hindering performance as well as driving up costs.
Another problem related to the client’s set-up for its own customers. Each customer had an individual public URL to access the CRM service, bringing additional costs and complexity to the monitoring of EKS clusters.
Expert cloud engineering skills were needed to resolve these issues. Sourced Group, an Amdocs company (Sourced), was appointed to take on the challenge.
“The database incompatibility issue was very unusual as Aurora should have been the ideal fit. Our AWS cloud engineering and database experts rallied to establish the root cause of the problem before figuring out how to fix it. Then we turned our attention to other matters that were preventing the client from enjoying the full benefits of AWS.”
Mike Lazenby, Lead Cloud Operations Engineer
Sourced Group
Solution: Tailored Approaches Optimise Costs and Performance
Our first step was to evaluate the incompatibility with Amazon Aurora and determine whether we needed to refactor or re-platform the client’s application. After an initial assessment, we opted to leverage MariaDB in a proof-of-concept deployment. This open-source relational database was created by the original developers of MySQL, and we used Amazon RDS to handle its set-up, operation, and scaling. RDS MariaDB significantly improved performance; it worked more effectively with the application’s query structure and brought costs down to an acceptable level.
Once the database issues were under control, we looked at the monitoring situation. Due to the high number of public-facing URLs that had to be monitored, AWS’ out-of-the-box solution was disproportionately expensive. To overcome this, we wrote a bespoke EKS monitoring solution that was sophisticated enough to meet the client’s needs while reducing costs by 99%. We set this up so the client can mute the solution when doing its own maintenance, so our managed services team isn’t alerted unnecessarily.
When penetration testing revealed that additional protection was needed, we introduced AWS Web Application Firewall (WAF). This service protects against common web exploits and bots that can affect availability, compromise security, or consume excessive resources. It enabled us to introduce the OWASP Top Ten ruleset in line with industry standards, customised according to roles as needed. We also introduced the WAF to the pre-prod environment to identify whether items scheduled for production would result in any problems. It became apparent that AWS’ managed rules for WAF defaulted to the latest updates which weren’t always compatible with the application. So, we configured this to happen automatically in pre-prod, but manually in production environments, ensuring updates only went ahead when they had been fully tested.
Outcome: Operational Improvements Unlock Capacity for Innovation
The infrastructure changes and solutions we implemented got the cloud strategy on track to achieve its goals, improving cost management, stability and reliability. This shaved thousands of GB pounds off monthly cloud costs, which made a significant difference to this scale-up business. We also introduced measures to mitigate the risk of future performance issues.
Collectively, this activity allowed the in-house team to dedicate more time and effort to continual improvement and product innovation. What’s more, strong foundations in the cloud gave the business leaders confidence to execute their plans knowing the IT infrastructure could cope with accelerated growth.
“When standard methods didn’t work, we found alternative ways to meet the client’s goals. A DevOps-led approach to cloud operations achieved better outcomes at a price that suited a scale-up business.”
Mike Lazenby, Lead Cloud Operations Engineer
Sourced Group
