Case Study: NVIDIA achieves real-time, scalable logging and analytics with Elastic

Nvidia - Customer Case Study

NVIDIA’s Kratos team built a data-platform-as-a-service to support GeForce NOW cloud gaming, facing a multi-tenant, high-throughput logging and analytics challenge: ingest real‑time and batch events at scale, maintain security and high availability, and give developers and analysts self‑service access to QoE, operations and business metrics. The platform needed to minimize engineering overhead for dashboards and BI while ensuring zero data loss and fast SLAs.

Kratos delivered an AWS-based pipeline (API Gateway → Kafka → ELK, with Spark/Presto and S3 for batch) that auto-scales to handle ~1M events/sec, uses VPC/LDAP for security, and provides HA and data replication. A Visual Logging extension (NGILogger) enabled zero‑engineering visualizations, automated dashboards and JDBC access for BI tools; the result was always‑on analytics with ~10s real‑time and 60s batch SLAs (96th percentile), faster troubleshooting, and broad self‑service reporting for engineers and analysts.

Nvidia

Satish Dandu

Data Science & Engineering Manager