October 2020 Summaries

Memgraph 1.2's release enhances compatibility with the graph ecosystem by introducing support for Bolt Protocol versions 4 and 4.1, facilitating seamless integration with existing Neo4j drivers across various programming languages. This development allows developers to interact with Memgraph using familiar tools and enhances the efficiency of querying Memgraph. The implementation of the Bolt protocol necessitated careful handling of message exchange, serialization, and chunking, as well as the management of multiple protocol versions to maintain backward compatibility. Key changes include the handling of PULL and DISCARD messages and the introduction of parameters like 'n' and 'qid' for more efficient data management. Additionally, Memgraph is developing its own drivers to optimize performance further, despite already supporting Neo4j drivers. This release underscores Memgraph's commitment to keeping pace with evolving protocol versions and improving developer experience through detailed protocol specifications and innovative solutions to challenges like memory management and transaction handling.

Memgraph's in-memory graph storage and analytics engine has undergone significant enhancements from versions v0.15.2 to v1.1.0, resulting in up to 50% better memory usage and improved throughput. The introduction of a new storage engine in v0.50.0 eliminated the need for whole copies of modified database objects, significantly reducing memory usage. Further improvements in v1.1.0 included the encoding and compression of node and edge properties, and the adoption of skip lists for indexing, contributing to better performance and near-linear scalability. Benchmarking showed that while certain queries like "Insert New Node" saw similar performance across versions, others benefited from the new optimizations, showcasing Memgraph's ability to handle hybrid transactional-analytical workloads efficiently. The enhancements allow storage of the same dataset in significantly less memory, especially when properties on edges are disabled, exemplifying Memgraph's advancements in providing a powerful graph analytics platform.