Success Story
N5 cuts the costs of a QA environment by 54% and gains full observability across its cloud accounts
At a glance
N5 Now is an Argentine software company that builds digitalization platforms for banks and insurers and operates in 18 countries. Craftech worked on the multi-customer infrastructure N5 runs on AWS and Azure. It implemented full observability in a banking customer's accounts, optimized costs and scaling, and migrated self-managed Kafka to Amazon MSK in development. An insurer's QA account cut its costs by 54%.
54%
lower costs in an insurer's QA account (Azure/AKS)
N5 Now is an Argentine software company founded in 2017 that builds digitalization platforms for banks and insurers, so they can modernize without replacing their legacy technology. It operates in 18 countries. Craftech worked on the infrastructure N5 runs for its customers on AWS and Azure.
Solution implemented
- Full observability with New Relic in the banking customer's development and production accounts (AWS): Kubernetes agent, APM with auto-instrumentation, FluentBit for logs, CloudWatch Metric Streams, Real User Monitoring, synthetic tests and dashboards
- Prior evaluation of several observability tools, with a cost analysis
- Cost optimization on Azure (AKS): tuning of requests/limits and of overprovisioned resources
- Node autoscaling with Karpenter
- Migration of self-managed Kafka to Amazon MSK in development
- Automated deployment and Git versioning of Amazon QuickSight dashboards
Business Challenge
N5 runs multi-customer infrastructure on AWS and Azure. A banking customer's account had no observability, which made it hard to detect and resolve incidents. Other customers' accounts had high costs due to overprovisioning and lack of autoscaling. In addition, stateful services inside Kubernetes (Kafka, Redis, RabbitMQ and MongoDB) caused recurring operational issues.
Solution
After evaluating several tools with a cost analysis, Craftech implemented full observability with New Relic in the banking customer's development and production accounts on AWS: Kubernetes agent, APM with auto-instrumentation, FluentBit for logs, CloudWatch Metric Streams, Real User Monitoring, synthetic tests and dashboards. On Azure (AKS) it optimized costs by tuning requests/limits and overprovisioned resources. It added node autoscaling with Karpenter, migrated self-managed Kafka to Amazon MSK in development and automated the deployment of Amazon QuickSight dashboards, with Git versioning.
Outcomes
- End-to-end observability (APM, logs, infrastructure, RUM and alerts) in the banking customer's development and production accounts
- 54% lower costs in an insurer's QA account (Azure/AKS)
- Kafka migrated to Amazon MSK in development, with no broker operations
- Node autoscaling with Karpenter
- QuickSight automated and versioned in Git
Stack & technologies
