LNX aims to be one of the most performant and cost-effective search engine alternatives to Elasticsearch and Algolia for typo-tolerant search. It offers a wide range of features thanks to the ecosystem it stands on, including but not limited to: complex query parser, typo-tolerant fuzzy queries, typo-tolerant fast-fuzzy queries (pre-computed spell correction), more-like-this queries. It thrives to provide more for your hardware. In this talk, we take a look at how LNX uses Tantivy and other optimisations to create a lightning-fast search experience for you and your users.

In today’s modern world, any systems/servers/applications produce an increasing tremendous amount of logs and you often have to choose between costly SaaS services or on-premise solutions that are hard to manage at a large scale.

That’s why we build Quickwit from the ground up to make it cloud-native. In this talk, we will see how we re-designed indexing and search to truly decouple storage and compute, and create stateless search instances while keeping a sub-second response time.

Cluster membership management is one of the critical aspects of a distributed system. To put it simply, nothing can work if nodes don’t know about each other and what service or resource each can offer.

In this talk, first, we briefly describe Quickwit’s previous cluster management implementation. Then deeply explore the new implementation based on the scuttlebutt algorithm while highlighting our motivations and what the current rust ecosystem offers. The algorithm presented here is also what powers Apache Cassandra cluster feature. We finally conclude by presenting a couple of tricks we used to overcome some implementation challenges while contrasting with the solutions adopted by Apache Cassandra.