How Imperva Transformed Its Data Infrastructure with Twingo and SingleStore

About Imperva

Imperva, a cybersecurity leader and now part of the Thales Group (as of December 2023), helps organizations protect critical applications, APIs, and data at scale. With a strong emphasis on application security, Imperva provides services to enterprises ranging from banks and large corporations to small businesses, ensuring comprehensive cloud-based protection. Their Cloud WAF (Web Application Firewall) safeguards customer traffic, while DDoS protection and other security solutions mitigate evolving cyber threats.

Challenges

Imperva faced significant scalability and performance limitations with its previous statistical engines, which evolved through three generations:

• First-generation: Built in-house, constrained by physical infrastructure.
• Second-generation: Based on Cassandra, but still required manual scaling and long development cycles (adding a new metric took at least two weeks).
• Third-generation (current): A cloud-native approach was needed to ensure:
  • Scalability – Rapid growth demanded a system that could handle exponential data expansion across multiple global regions.
  • Performance – Queries needed to be sub-second to support customer dashboards and billing.
  • Flexibility – The ability to quickly integrate new metrics (e.g., for emerging cyber threats) was crucial.
  • Reliability – The previous system had manual dependencies, leading to frequent downtime and operational inconsistencies.

Imperva required a fully managed, cloud-based solution that not only met these requirements but also ensured that all data was injected in near real time and supported very high concurrent customer-facing transactions.

Solution

Imperva partnered with Twingo to modernize its data infrastructure. During an extensive Proof of Concept (POC), two main alternatives— Druid Imply and SingleStore—were evaluated. SingleStore emerged as the clear winner, delivering:

• Seamless Scalability: Capable of processing 95% of queries in under 400 milliseconds across 40TB of data.
• Real-Time Performance: Essential for dashboards, analytics, and billing in a highly concurrent, customer-facing SaaS environment.
• High Availability: Clusters deployed on AWS Multi AZ (Availability Zones) ensure robust uptime and resilience.

Twingo’s engagement went far beyond just system selection. Our experts designed the overall data schema and determined optimal sizing strategies, implemented advanced performance tuning and query optimization, and architected a highly available multi-AZ solution—all while leveraging a Kubernetes-managed environment to orchestrate containerized microservices. This comprehensive approach eliminated manual scaling challenges and significantly accelerated development cycles; adding a new metric was reduced from two weeks to just 30 minutes.

Results

The transformation led to significant improvements in Imperva’s data architecture:

• Scalability & Global Expansion:
  • Grew from 1 region to 4 to meet global compliance requirements.
  • Units expanded from 6 to 30, handling a vastly larger dataset.
• Unmatched Performance & Efficiency
  • P95% of queries execute in under 400 milliseconds, even with 40TB of data.
  • Adding new cyber threat insights for customers now takes just 30 minutes, compared to two weeks previously.
• Enhanced Stability & Reliability
  • More automated management reduces manual intervention and downtime, ensuring higher availability and consistent operations.
• Cost & Operational Efficiency
  • Fully cloud-native, eliminating manual infrastructure scaling challenges.
  • Reduced operational complexity, enabling Imperva’s teams to focus on cybersecurity innovation rather than maintaining data systems.

With Twingo and SingleStore, Imperva successfully transitioned to a scalable, real-time, cloud-based data solution, improving security insights, customer experience, and operational agility.