Case study
Search Twice as Fast for an Asian Matchmaking Platform's 35 Million Users
How SquareShift benchmarked and reconfigured a 47-node Elasticsearch cluster with ES Rally, more than doubling query performance for a matchmaking platform serving 35 million users.
Book a session →
Client
A large matchmaking platform in Asia that serves more than 35 million users through its web and mobile apps.
Industry
Technology
Services
Stack
Elasticsearch · Amazon Web Services (AWS) · Google Cloud Platform · Elasticsearch Rally (ES Rally)
How did SquareShift more than double query performance on a 47-node cluster?
The challenge
After the client moved its Elasticsearch cluster from AWS to Google Cloud, query performance slowed down for a matchmaking platform serving more than 35 million users through web and mobile apps.
What we delivered
SquareShift compared the AWS and GCP setups, then ran cluster analysis and benchmark tests with ES Rally. The team reconfigured all 47 nodes and improved cluster layout and indexing patterns to match how Google Cloud actually performs.
The outcome
Elasticsearch query latency improved by more than 2x, and the cluster now sustains 90,000 requests per second. For moves between platforms and the tuning after them, see our Elasticsearch migration services.
A lift-and-shift migration copies the infrastructure, not the tuning — a cluster configured for one cloud's disk and network profile needs reconfiguring, not just relocating, to perform the same way on another.
Platform & Software Engineering Practice Lead, SquareShift
