For measuring the latency of requests to the RPC endpoint, you can save and run the following script, which will send 20 "getSlot" requests sequentially, while keeping HTTP connection open, like in most of programming language libraries.
test_latency.sh
#!/usr/bin/env bash
API_KEY="YOUR-API-KEY"
RPC="https://solana-rpc.rpcfast.net/?api_key=${API_KEY}"
temp=$(mktemp)
trap "{ rm -f $temp; }" INT TERM EXIT
unset RPCS; declare -a RPCS
unset OPTS; declare -a OPTS
for i in {1..20}; do RPCS+="$RPC "; OPTS+="-o /dev/null "; done
curl ${RPCS[@]} ${OPTS[@]} -s -w '%{time_total}\n' -d '
{
"jsonrpc":"2.0",
"method":"getSlot",
"id":1
}' >> $temp
cat $temp | awk '{ printf "Request #" NR ": %.2f ms\n", $1 * 1000; sum += $1 } END { if (NR > 0) printf "=====================\nAverage: %.2f ms\n", (sum / NR) * 1000 }'
