Skip to content

CrossDB vs STL Map Benchmark

Test tool: CrossBench

DB Driver: STLMap CrossDB

Test Config: Random Access, Single Thread, Bind CPU Core, To make test fair, STL map uses read-write lock as CrossDB is thread-safe with read-write lock also.

Test Server

CPU         : Intel(R) Xeon(R) Gold 5318Y CPU @ 2.10GHz cache size 36864 KB
HDD         : DELL PERC H755 Front SCSI Disk
OS          : Ubuntu 20.04
STL         : GCC v10.2.1
CrossDB     : v0.5.0

In-Memory Database Benchmark

Test Script

loop="1 2 3"
./crossdb-bench.bin -H -r 0
for i in $loop; do ./crossdb-bench.bin -s m -i 1k   -q 40m -u 30m -Q -H -c $cpu; done
for i in $loop; do ./crossdb-bench.bin -s m -i 10k  -q 40m -u 30m -Q -H -c $cpu; done
for i in $loop; do ./crossdb-bench.bin -s m -i 100k -q 40m -u 30m -Q -H -c $cpu; done
for i in $loop; do ./crossdb-bench.bin -s m -i 1m   -q 10m -u 10m -Q -H -c $cpu; done
for i in $loop; do ./crossdb-bench.bin -s m -i 10m  -q 10m -u 10m -Q -H -c $cpu; done
for i in $loop; do ./crossdb-bench.bin -s m -i 100m -q 10m -u 10m -Q -H -c $cpu; done

loop="1 2 3"
./stlmap-bench.bin -H -r 0
for i in $loop; do ./stlmap-bench.bin -s m -i 1k   -q 40m -u 40m -Q -H -c $cpu; done
for i in $loop; do ./stlmap-bench.bin -s m -i 10k  -q 20m -u 20m -Q -H -c $cpu; done
for i in $loop; do ./stlmap-bench.bin -s m -i 100k -q 10m -u 10m -Q -H -c $cpu; done
for i in $loop; do ./stlmap-bench.bin -s m -i 1m   -q 5m  -u 5m  -Q -H -c $cpu; done
for i in $loop; do ./stlmap-bench.bin -s m -i 10m  -q 2m  -u 2m  -Q -H -c $cpu; done
for i in $loop; do ./stlmap-bench.bin -s m -i 100m -q 1m  -u 1m  -Q -H -c $cpu; done

Small Data Set Test

Large Data Set Test

Comments