According to the documentation here: TiDB Introduction | PingCAP Docs, ‘The computing layer supports a maximum of 512 nodes, each node supports a maximum of 1,000 concurrencies’. Does this mean that a TiDB cluster currently supports up to 512 x 10^3 tps(transactions per second)? If the answer is yes, is there any benchmark results that approximates it?
Eric Kaufmann
Currently, the largest TiDB cluster with benchmark result I can find is here: Scaling TiDB To 1 Million QPS. A Flipkart journey | by Sachin Japate | Flipkart Tech Blog. It is a cluster with 12 instances of TiKV nodes. It seems it already reaches a throughput of 1M QPS for read loads. I can’t imagine the throughput of a monster cluster with 512 TiKV nodes. It should be much higher than 512x10^3, right? Or have I interpreted those numbers incorrectly? Somebody can give me a hint here?