In this post we take a closer look at the results from the Elastic blog post benchmarking Elasticsearch vs. OpenSearch. We discover that inlining has a large impact on the performance, and use that to make a small Lucene change that makes some queries almost twice as fast.
In this post we take a closer look at the results from the Elastic blog post benchmarking Elasticsearch vs. OpenSearch. We discover that no real conclusions can be drawn from that benchmark, due to issues with queries matching no documents, caching and latency.
In this blog post we dig into an outlier query from benchmarking Elasticsearch vs. OpenSearch, and discover how one index setting can make text queries both 490 times faster (by not scoring documents) and 1875 times slower (to save disk).
In this post we learn that a significant part of the time indexing logs into Elasticsearch can be spent parsing dates, and explore making it faster.
In this post we deep-dive into exactly what Lucene, the popular search index library, tests in it's nightly benchmark.
What is really the performance impact of an int vs. an Integer? The answer turns out to be closely related to CPU caches.
Blunders does of course use Blunders to become faster. In this post we dig into how that is done.