Rakbank’s retail digital banking platform leverages a centralized observability solution built on Amazon OpenSearch Service that handles 150 GB of daily logs through a hot, UltraWarm, and cold data tiering model. This approach supports stringent seven-year regulatory retention and delivers roughly 55% cost savings compared to legacy architectures.
- Centralized log ingestion processes 150 GB daily with multi-tier OpenSearch storage
- Achieves ~55% cost savings via hot/UltraWarm/cold tiering and S3 archival
- Cross-account architecture isolated with Transit Gateway and fine-grained access control
Infrastructure signal
Rakbank’s observability infrastructure targets large-scale log ingestion at roughly 150 GB per day, derived from containerized microservices running on Amazon EKS. The platform employs a multi-AZ Amazon OpenSearch Service domain that administrates data across hot, UltraWarm, and cold storage tiers, thereby optimizing compute and storage costs across usage patterns. Logs are immutable append-only documents, which simplifies data management and allows for effective rolling of indices before tier migration.
The design spans two AWS accounts to isolate application workloads from observability operations and uses a private AWS Transit Gateway for secure cross-account log transfer, ensuring no data exits the AWS network. OpenSearch clusters in the observability account are distributed across three zones with dedicated cluster manager and data nodes for availability and horizontal scaling. Complementing this, archived log data complies with a seven-year retention policy mandated by regulatory requirements, leveraging Amazon S3 archival storage.
Developer impact
Developers benefit from a streamlined, standardized log collection mechanism via Fluent Bit, which is deployed as a DaemonSet across EKS pods. This uniform instrumentation approach reduces complexity across retail, business, and wholesale banking domains. Logs are written only once to the OpenSearch indices, eliminating overhead from updates or deletions and improving ingestion efficiency.
Access to OpenSearch Service Dashboards is integrated with Rakbank’s enterprise identity framework through SAML 2.0 single sign-on, allowing developers and analysts to securely visualize observability data without managing separate credentials. The separation of ingestion and user access paths enables granular security controls and eases troubleshooting responsibilities, aligning developer workflows with compliance and security requirements.
What teams should watch
Cloud infrastructure and observability teams need to monitor the cross-account permissions and network configurations that govern secure log flow through the Transit Gateway and enforce strict VPC-level access controls. As the platform scales, teams should validate index rollover and tier migration processes for consistency and capacity management, particularly as log volumes grow beyond current baselines.
Development and compliance teams must keep an eye on archival policies to assure retention windows meet evolving regulatory demands, particularly the seven-year log retention for Central Bank of the UAE compliance. Observability teams should also track dashboard usage patterns and query performance to optimize user experience and potentially tune the hot tier cluster resources accordingly.