Background:
CyberAuctions.com is a complete on-demand auction system for businesses and non-profits. They provide a convenient and efficient bidding system for both sellers and buyers.
Challenge:
Auctions hit peak traffic near the end as buyers compete to outbid each other before the close. Application-database response times are crucial as bidding promotes more bidding which results in higher revenues for CyberAuctions.com. Finding a solution that identifies the SQL bottlenecks and improving performance was directly tied to auction revenues.
Solution:
We chose Amazon Aurora for MySQL for its self-management, ease of horizontal scaling, and roll back capabilities. Our deployment had two application, EC2 instances connected to three Aurora backend servers (one write, two read replicas). Heimdall Data’s transparent deployment complemented Aurora with:
- Better use of read replicas (aka Read/Write splits): Without any application code changes, Heimdall routed the queries to the write servers and to the appropriate read replicas enabling horizontal scale out. Additionally, the replication lag detection feature ensured that any Aurora read data would be fresh and up to date.
- SQL caching: We choose Elasticache for Redis for our caching grid. The best part was we didn’t have to do any code changes to our application to take advantage of caching. We just turned Heimdall’s Auto-tune feature and it worked. Caching and invalidation was automated.
- SQL Analytics: Ability to identify performance bottlenecks and resolve them quickly
Customer Benefit:
Heimdall Data’s reporting system provided insights into how we could improve our SQL performance. This alone made the product very useful. We are an online auction company and caching improved our website response times up to 50 percent. We are very pleased with the performance gains and visibility delivered from the Aurora and Heimdall Data joint solution.
Resources and links:
