May 2017 Summaries

Heap, a customer analytics tool, significantly improved its Postgres cluster's CPU efficiency by applying basic performance analysis techniques, leading to a tenfold enhancement and substantial cost savings. Initially, Heap's unique indexing strategy, which utilized Postgres' partial indexing for faster event retrieval, resulted in unexpectedly high CPU usage. This was attributed to the high number of regular expression evaluations required for each event insertion. However, further investigation using flame graphs revealed that the CPU load was due to frequent fetching and parsing of index metadata rather than predicate evaluations. By implementing batch inserts instead of individual ones, Heap reduced the need for repeated metadata processing, achieving a 10x reduction in CPU usage. This simple change not only improved ingestion throughput by approximately tenfold but also eliminated the need for additional CPU resources in the foreseeable future.

Heap has announced a $27 million Series B funding round co-led by NEA and Menlo Ventures, with continued support from Initialized Capital and Pear VC. While Heap does not internally celebrate fundraising, the company is committed to using the funds to enhance customer satisfaction through automation and team growth. Heap aims to reduce the complexities of analytics by automatically capturing all user data, eliminating the need for manual event definitions and tracking code. The company is also focused on expanding its lean team, which currently operates with minimal staff but prioritizes autonomy and efficiency. Heap's ultimate goal is to make data more accessible and valuable for everyone, and it expresses gratitude to its 6,000+ customers for their support as the company works to further revolutionize data analytics.