Lindorm helps Napatech significantly improve storage efficiency and indexing performance — enabling packet metadata collected at over 100 Gbit/s per source to be stored, indexed, and analyzed while reducing O&M overhead.
Customer profile
Napatech (OSE: NAPA.OL) is a Denmark-based company founded in 2003. It was the first to apply field-programmable gate array (FPGA) technology to network management and security applications. Napatech serves customers in the telecommunications, finance, and internet industries across the United States, Asia Pacific, Europe, the Middle East, and Africa — helping them reconstruct computing platforms with FPGA to deliver secure, high-performance network data processing.
Business challenges
As Napatech's customers scaled data collection to 200 Gbit/s, three problems emerged:
Storage and indexing bottlenecks: Upper-layer software generated growing volumes of packet metadata that overwhelmed existing open source HBase and Elasticsearch backends — both in write throughput and query performance.
Rising costs and O&M burden: Scaling out with additional nodes drove up infrastructure costs and added significant operational complexity.
Urgent need to improve system performance: The performance of the storage system and the indexing system for the metadata of the data packets urgently needed to be improved.
Solution
Napatech adopted Lindorm, a cloud-native multi-model database service built on a compute-storage decoupled architecture. This architecture allows Napatech to scale storage and compute independently, keeping costs predictable as data volumes grow.
The deployment handles packet metadata ingested from multiple data sources, each at over 100 Gbit/s. Lindorm supports both full data imports and incremental data channels for real-time ingestion, and integrates with big data platforms to enable offline analysis on historical data.
Benefits
Lindorm directly addressed each of Napatech's core challenges:
Storage and indexing bottlenecks resolved: Lindorm stores and indexes packet metadata from multiple data sources at over 100 Gbit/s per source, and supports backtracking analysis on the full dataset.
Costs and O&M complexity reduced: Lindorm's elastic scalability and simplified database architecture eliminated the need to maintain separate HBase and Elasticsearch clusters, reducing both infrastructure spend and operational overhead.
Unified data platform established: Lindorm consolidates real-time ingestion, batch import, and big data integration into a single system, removing the need for multiple specialized backends.