Background:
After Questis completed a major software release and was preparing to onboard a large customer, it was time to focus on improving the performance and scalability of the application. The approach was first, to develop industry differentiating functionality and then, to optimize for performance.
Challenge:
The Questis team realized that their backend infrastructure required additional development work to improve performance, provide high-availability, and scalability.
Solution:
Questis chose Amazon Aurora for Postgres for its:
- Self-management
- Excellent sub-second replication times
- Auto-scaling of database storage
The deployment had three application EC2 instances running under ECS connected to two Aurora backend servers (one write, one read replica). Heimdall Data’s transparent deployment complemented Aurora with:
- Read/Write query separation: Without any application code changes, Heimdall routed the queries to the write servers and to the appropriate read replicas. Replication lag detection ensured that data retrieved was never stale whether from cache or from Aurora.
- Query caching: Questis chose Elasticache for Redis as part of their caching grid. No code changes to the application were required to take advantage of caching. Heimdall’s proxy provided both the caching and invalidation logic. In one-click, query caching was enabled.
- Automated Failover: This added the ability to track and failover the database without requiring an application restart.
- SQL Analytics: The Heidmall Central Console provided client-side and server-side analytics to identify performance bottlenecks and resolve them quickly. Check out the below categories that give useful statistics:
Customer Benefit:
Heimdall proxy provided insights into how Questis could optimize slow SQL queries. This alone made the product very useful. Query caching improved website response times up to 46 percent. Questis was very pleased with the performance gains and visibility delivered from the Aurora and Heimdall Data joint solution.
Resources and links:
