Lilith Games cut report query latency from 15 seconds to 1.5 seconds — a 5–10x analytics performance improvement — after migrating to AnalyticDB for MySQL for real-time game ad measurement.
The challenge
In the game industry, rising Internet traffic costs push companies toward fine-grained ad delivery optimization. Lilith Games spends tens of thousands of dollars per minute on promotions, so every ad campaign decision must be backed by real-time data.
Their measurement system had to handle wide tables with over 300 columns of multidimensional metric data — while supporting elastic scaling under heavy query loads without performance degradation. Legacy infrastructure couldn't deliver second-level query results at that scale and column count.
Solution
Lilith Games built their next-generation real-time game advertising platform on AnalyticDB for MySQL, which addresses each of their core requirements:
-
Second-level query performance. Vectorization and inverted indexes deliver query results in seconds, giving ad operations teams real-time visibility into campaign performance without waiting for batch reports.
-
Wide tables with thousands of columns. PAX-based hybrid row-column storage optimizes I/O for wide-table workloads. Optimized memory control resolves I/O amplification issues when updating wide tables.
-
On-demand elasticity. Resources automatically scale based on concurrent and queued query volume, so performance stays consistent during peak campaign periods.
Results
The migration delivered measurable gains across ad operations:
-
5–10x analytics performance improvement
-
Report query latency dropped from 15s to 1.5s
-
Faster ad delivery decisions in user acquisition campaigns