Nasstar migrates RDG’s LSM database to a new cloud-native architecture
Nasstar migrates RDG's legacy database to a new cloud-native digital architecture
Rail Delivery Group (RDG) brings together the companies that run Britain’s railway into a single team with one goal; to deliver a better railway for passengers and the wider community. All the UK’s passenger and freight rail companies are members of the RDG, as well as Network Rail and HS2.
RDG provides the services and support that enables their members to succeed in transforming and delivering a thriving rail service, benefitting customers, taxpayers, and the overall economy. They also act as a voice for freight and passenger operators in the industry and deliver important national ticketing, information and reservation services for passengers and staff on behalf of member companies.
Amazon Web Services
Amazon Aurora (Database-as-a-Service)
AWS Database Migration Service
RDG’s Live Sales Management (LSM) MySQL database has been hosted on AWS Elastic Computing (EC2) instances for a number of years. While the solution has performed well during this time, several technical constraints were identified on the event of a database failure. Significant support overheads were also present as Nasstar was required to self-manage a MySQL database on a set of EC2 instances.
An opportunity was identified to bring the LSM database up to date with the latest technologies and improve aspects of the existing solution, including its replication and failover capabilities. This would involve migrating the existing LSM database to the Amazon Aurora service to leverage the many benefits of a cloud-native Database-as-a-Service (DBaaS) solution.
Following a successful migration to the newly provisioned Amazon Aurora service in November 2021, RDG has realised several business benefits directly attributable to the migration of the LSM database. These include:
The migration to the cloud-native Amazon Aurora service mitigated the risks associated with the previous replication and failover capability. Amazon Aurora allows for the provisioning of managed database backups and routine patching/maintenance schedules. Additionally, any existing technical debt has been removed.
The Amazon Aurora service has reduced the outage time in the event of a database failure from around 30 minutes to approximately 80 seconds.
Leveraging the benefits of cloud-native services has reduced the operational overhead costs associated with the previous approach of a self-managed solution hosted on EC2.
The new, modern relational database service offers high performance at scale, improving on the performance reported with the previous solution.
Amazon Aurora’s CloudWatch metrics and alarms have provided enhanced database monitoring and alerting capabilities. Performance Insights are enabled on the production database to further expand on the monitoring features. This helps RDG to further illustrate and help analyse cluster performance.
To improve the service of the LSM database, Nasstar conducted a full assessment of available database migration approaches to determine which solution would best fit RDG. AWS Database Migration Service (AWS DMS) was selected as the secure and efficient utility to support the migration of the LSM database from EC2 into Amazon Aurora.
A number of key factors were considered to support this migration, including:
• Service outage requirements
• Database volume sizes
• Migration complexity and viability (primarily determined by outage requirements and data sizes)
• Data validation and consistency checks (significant effort must be made to ensure the metadata, i.e., the data model and datasets are fully aligned during all phases of the migration to ensure data integrity)
Using the AWS DMS service, Nasstar utilised the ’pre-migration assessment,’ a valuable tool to provide analysis that warns of any issues before the migration was carried out. Assessments were carried out during the early analysis and design phases providing crucial insight into the related migration steps, limitations and any configuration changes that would be required.
Following the assessments and subsequent reviews, Nasstar determined that AWS DMS was a viable solution, and any findings were considered as part of the overall migration solution.
The migration of the LSM database was rehearsed in multiple test environments prior to production, enabling Nasstar to carry out testing as follows:
• proving of the required Infrastructure-as-Code deployments
• validation of the migration approach using DMS
• data integrity checking post-migration
• thorough real-world load testing
The load testing provided confidence that the target Amazon Aurora service could sustain both normal and peak loads, and that there would not be any performance impacts of the new approach.
RDG and Nasstar have engaged regularly over the last four years about re-scaling the LSM service. Due to a requirement for the reduction in the utilisation of the service and with an appetite for the rail industry to remove the use of paper ticketing over a period of time, Nasstar provided RDG a number of key opportunities that would enhance efficiency, cost, and performance - one of which was the database migration for an AWS Aurora Database. Since the migration, we have noticed cost savings and a 10% transactional performance improvement. Together with this change being an enabler for other service improvements, it also future proofs the RDG environment.