Introducing Husky, Datadog's Third-Generation Event Store

This article discusses the development of "Husky", an event storage system built by Datadog. It explains how the company transitioned from storing pre-aggregated timeseries data to handling logs, which require a different approach due to their larger size and high cardinality requirements. The initial version of Datadog's Logs system had limitations in terms of reliability and scalability, leading to the development of a new architecture that separated storage from clustering. However, as the company's platform grew, more issues emerged, necessitating the creation of an entirely new storage system called "Husky". This new system separates compute from storage, allowing for independent scaling of ingestion, storage, and query paths. It also provides flexibility in terms of isolation and performance tuning. The article concludes by noting that this is just the first part of a series discussing Husky's storage system in more detail.