Elasticsearch Adoption and Architecture Cases in Major Chinese Companies

Elasticsearch has become a core component for many Chinese internet companies, including JD.com, Ctrip, Quark, 58.com, and Didi, enabling fast search, real‑time data analysis, log monitoring, and security analytics across massive structured and unstructured datasets.

JD.com Order Center Evolution – To handle billions of order documents and hundreds of millions of daily queries, JD.com migrated order queries from MySQL to Elasticsearch, adopting a real‑time active‑standby cluster with VIP load balancing, client gateway nodes, and data nodes with one primary and two replicas. The team tuned shard numbers to balance single‑ID lookup throughput and aggregation performance.

Ctrip Use Cases – Ctrip deployed Elasticsearch for hotel order search, building a real‑time index on sharded databases and exposing a dedicated web service for queries. They also described large‑scale cluster management, handling 600 billion daily index entries, peak write rates of millions of records per second, and a cluster of over 120 data nodes across 70 servers.

Quark Order Center – Facing daily order volumes of up to one million, Quark replaced a hot‑table sharding approach with an Elasticsearch cluster to store searchable fields while keeping detailed order data in MySQL. The solution uses eight primary shards per index, storing 140 million documents per index and scaling to 64 GB of disk usage per node.

58.com Information Security – The Elastic Stack (Elasticsearch, Kibana, Logstash, Beats) is used for security event storage, high‑throughput search, low‑latency optimization, and Kibana visualizations, supporting security monitoring and incident response.

Didi Multi‑Cluster Architecture – Didi operates more than 3 500 Elasticsearch instances with over 5 PB of data. Data ingestion is managed by a Kafka‑based Sink service that streams logs, binlogs, and custom metrics into Elasticsearch, while a Gateway service provides HTTP, TCP, and SQL interfaces, load balancing, rate limiting, and multi‑cluster disaster recovery.

Practical Order Search Solution – A service‑oriented architecture wraps Elasticsearch and database shards behind a unified order service API, enabling front‑end and back‑end applications to query orders efficiently. Visualizations and dashboards are built with Kibana, and the system supports real‑time updates and high query concurrency.

Overall, the article demonstrates how Elasticsearch can be scaled to billions of documents, support high query throughput, and integrate with streaming pipelines to meet the demanding search and analytics needs of large‑scale internet services.