February 2025 Summaries

Harper offers a streamlined solution for application delivery by consolidating various technologies into a single, high-performance, container-friendly platform, allowing developers to focus on innovation rather than orchestration. This approach results in significant latency reductions and cost savings, as demonstrated by Edison Interactive's dramatic improvements in response times and infrastructure expenses. Harper supports diverse use cases, including data enrichment and personalization, GraphQL caching, fraud detection, IoT processing, and inventory management, by leveraging its distributed low-latency data processing capabilities. The system's flexible architecture and direct communication protocols make it ideal for IoT networks, while its ability to synchronize inventory data in real-time ensures efficient management. By adopting Harper, businesses can gain a strategic advantage through enhanced performance and user experience across their application delivery systems.

In Harper 4.4, the introduction of the replication system "Plexus" has significantly enhanced performance, security, and reliability by eliminating the need for a message broker and establishing direct connections between nodes, enabling optimizations to the TCP level, secure mTLS connections, and robust consistency tracking. The new system demonstrated a remarkable average 50% increase in replicated write performance over Harper 4.3, with tests conducted across clusters of 4, 5, 6, and 8 nodes using both random and sequential primary keys to assess variability. The tests, which did not enable sharding, showed a mean performance improvement of 51% after excluding an outlier, indicating potential higher throughput with sharding. The benchmarks were conducted on a multi-node Harper cluster using Akamai Connected Cloud Instances, with high concurrency loads of 500 virtual users, testing record sizes of 400 bytes across 8 fields, and focusing on newly inserted records replicated to all nodes via REST API, executed by the k6 load test runner.

Harper introduces a pioneering backend system designed to enhance digital commerce by integrating data, application, streaming, and in-memory systems into a single, high-performance technology. This system is engineered to deliver exceptional speed, with capabilities to respond to client requests in as little as 300 microseconds and ensure sub-50 millisecond roundtrip latency worldwide, significantly improving page load times and addressing network latency challenges. Harper aids businesses in implementing content optimization strategies such as A/B testing, dynamic pricing, and personalization by processing data at the edge, thus minimizing the need for extensive backend interactions and enabling faster response times. With its core data engine outperforming traditional systems like MongoDB, Harper offers companies the ability to scale efficiently while integrating complex features without the need for additional technologies. By positioning itself as a transformative solution in the competitive digital commerce landscape, Harper empowers businesses to achieve both performance and agility, ultimately driving increased revenue and operational efficiency.

HarperDB has been recognized as an IDC Innovator in the report "IDC Innovators: Edge Inference Delivery, 2024" for its innovative approach to managing data-dependent workloads in highly distributed environments. The company aims to simplify and enhance the scalability of modern applications by integrating database caching, application, and messaging functions into a single, high-performance technology stack on one server. This architecture supports distributed systems and low-latency operations, enabling rapid data synchronization and decision-making across different locations. HarperDB’s system is designed to handle capacity limitations typical of edge resources, offering nearly limitless horizontal scaling and fault-tolerant data streaming, which has been successfully applied in projects such as the U.S. Navy's Common Operational Database. The company’s backend technology delivers exceptional performance improvements, contributing to significant business agility and operational resiliency for its clients.

Since November 2022, the development of large language model chatbots has demonstrated significant advancements in neural network research, which seeks to replicate human brain functions using mathematical models of neurons. These chatbots, capable of conversing convincingly, rely heavily on a technique called Retrieval Augmented Generation (RAG) to retrieve relevant information and enhance context for more accurate responses. The efficiency of RAG is challenged by the latency in retrieving data, especially from large, distributed sites. Traditional caching methods are inadequate due to the unique nature of each chatbot interaction, but an innovative approach called Noetic Caching offers a solution by caching relevant contextual data, allowing for quick and precise chatbot responses. This method, integrated with systems like Harper, provides a balance of speed, accuracy, and cost-effectiveness, making it a potential game-changer for reducing latency in generative AI applications.