diff --git a/README.md b/README.md index 7c9578587..4fcb863d6 100644 --- a/README.md +++ b/README.md @@ -175,19 +175,19 @@ Now that your MeiliSearch server is up and running, you can learn more about how ## Performances -When processing a dataset composed of _100 353_ documents with _352_ attributes each and _3_ of them indexed, which means more than _300 000_ fields indexed for _35 million_ stored, MeiliSearch is able to carry out more than _2.8k req/sec_ with an average response time of _9 ms_ on an Intel i7-7700 (8) @ 4.2GHz. +When processing a dataset composed of 5M books composed of a title and an author name, MeiliSearch is able to carry out more than 553 req/sec with an average response time of 21 ms on an Intel i7-7700 (8) @ 4.2GHz. Requests are made using [wrk](https://github.com/wg/wrk) and scripted to simulate real users' queries. ``` -Running 10s test @ http://localhost:2230 - 2 threads and 25 connections +Running 10s test @ http://1.2.3.4:7700 + 2 threads and 10 connections Thread Stats Avg Stdev Max +/- Stdev - Latency 9.52ms 7.61ms 99.25ms 84.58% - Req/Sec 1.41k 119.11 1.78k 64.50% - 28080 requests in 10.01s, 7.42MB read -Requests/sec: 2806.46 -Transfer/sec: 759.17KB + Latency 21.45ms 15.64ms 214.10ms 85.95% + Req/Sec 256.48 37.66 330.00 69.50% + 5132 requests in 10.05s, 2.31MB read +Requests/sec: 510.46 +Transfer/sec: 234.77KB ``` We also indexed a dataset containing about _12 millions_ cities names in _24 minutes_ on a _8 cores_, _64 GB of RAM_, and a _300 GB NMVe_ SSD machine.