Stress Test New VMware Infrastructure
I was performing some stress tests on Linux VMware machines for a new environment to evaluate and find the resource bottlenecks and how SAN storage reacts in terms of CPU I/O waiting before using this environment for production purpose.
I got the interesting results and would like to share with you guys.
The "stress" tool has been used with options shown below:
# stress -c 65 -i 60 -m 25 --vm-bytes 256M -t 120m -v
-c: number of CPU workers
-i: number of IO workers
-m: number of Memory workers
--vm-bytes: allocate memory size for each memory worker(default 256MB)
-t: the duration of running the test
-v: make the test in verbose mode
With this options, I went through 150 for 5min CPU load and almost 100 for 15min load which is pretty amazing.
It has carried out in a test VM with 2 vCPU and 4GB memory.
# stress -c 65 -i 60 -m 25 --vm-bytes 256M -t 120m -v
top - 17:44:56 up 55 min, 3 users, load average: 151.21, 124.73, 99.78
Tasks: 254 total, 67 running, 187 sleeping, 0 stopped, 0 zombie
Cpu(s): 74.5%us, 24.2%sy, 0.0%ni, 0.0%id, 0.0%wa, 0.0%hi, 1.3%si, 0.0%st
Mem: 3924688k total, 3837772k used, 86916k free, 260k buffers
Swap: 3354616k total, 2968112k used, 386504k free, 4040k cached
PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND
3309 root 20 0 6516 124 104 R 3.2 0.0 0:06.40 stress
3327 root 20 0 6516 124 104 R 3.2 0.0 0:06.40 stress
3330 root 20 0 6516 124 104 R 3.2 0.0 0:06.40 stress
3342 root 20 0 6516 124 104 R 3.2 0.0 0:06.40 stress
3372 root 20 0 6516 124 104 R 3.2 0.0 0:06.41 stress
3385 root 20 0 6516 124 104 R 3.2 0.0 0:06.41 stress
3387 root 20 0 6516 124 104 R 3.2 0.0 0:06.42 stress
3391 root 20 0 6516 124 104 R 3.2 0.0 0:06.41 stress
3393 root 20 0 6516 124 104 R 3.2 0.0 0:06.41 stress
3395 root 20 0 6516 124 104 R 3.2 0.0 0:08.21 stress
3403 root 20 0 6516 124 104 R 3.2 0.0 0:06.41 stress
3405 root 20 0 6516 124 104 R 3.2 0.0 0:06.41 stress
3409 root 20 0 6516 124 104 R 3.2 0.0 0:06.41 stress
3413 root 20 0 6516 124 104 R 3.2 0.0 0:06.41 stress
3417 root 20 0 6516 124 104 R 3.2 0.0 0:06.41 stress
3437 root 20 0 6516 124 104 R 3.2 0.0 0:06.41 stress
3315 root 20 0 6516 124 104 R 2.8 0.0 0:06.40 stress
3324 root 20 0 6516 124 104 R 2.8 0.0 0:06.39 stress
3333 root 20 0 6516 124 104 R 2.8 0.0 0:06.40 stress
3336 root 20 0 6516 124 104 R 2.8 0.0 0:06.39 stress
3339 root 20 0 6516 124 104 R 2.8 0.0 0:06.39 stress
3354 root 20 0 6516 124 104 R 2.8 0.0 0:08.21 stress
3381 root 20 0 6516 124 104 R 2.8 0.0 0:06.40 stress
3389 root 20 0 6516 124 104 R 2.8 0.0 0:06.41 stress
3397 root 20 0 6516 124 104 R 2.8 0.0 0:06.41 stress
3401 root 20 0 6516 124 104 R 2.8 0.0 0:06.40 stress
3407 root 20 0 6516 124 104 R 2.8 0.0 0:06.41 stress
3411 root 20 0 6516 124 104 R 2.8 0.0 0:06.41 stress
3415 root 20 0 6516 124 104 R 2.8 0.0 0:06.41 stress
3429 root 20 0 6516 124 104 R 2.8 0.0 0:06.41 stress
3447 root 20 0 6516 124 104 R 2.8 0.0 0:06.40 stress
3449 root 20 0 6516 124 104 R 2.8 0.0 0:06.40 stress
3453 root 20 0 6516 124 104 R 2.8 0.0 0:06.41 stress
3454 root 20 0 6516 124 104 R 2.8 0.0 0:06.61 stress
39 root 20 0 0 0 0 D 1.6 0.0 0:23.33 kswapd0
3306 root 20 0 6516 124 104 R 1.6 0.0 0:06.74 stress
3312 root 20 0 6516 124 104 R 1.6 0.0 0:06.74 stress
3318 root 20 0 6516 124 104 R 1.6 0.0 0:06.74 stress
3321 root 20 0 6516 124 104 R 1.6 0.0 0:06.74 stress
Following demonestrate the results of dd writing of 1GB to the SAN with different block sizes of 1MB, 10MB and 100MB.
The bigger the block size get, the slower the writing to the SAN becomes:
# dd if=/dev/zero of=/tmp/test22 bs=1M count=1000
1000+0 records in
1000+0 records out
1048576000 bytes (1.0 GB) copied, 77.165 s, 13.6 MB/s
# dd if=/dev/zero of=/tmp/test22 bs=10M count=100
100+0 records in
100+0 records out
1048576000 bytes (1.0 GB) copied, 104.555 s, 10.0 MB/s
# dd if=/dev/zero of=/tmp/test22 bs=100M count=10
10+0 records in
10+0 records out
1048576000 bytes (1.0 GB) copied, 1204.23 s, 871 kB/s
I got the interesting results and would like to share with you guys.
The "stress" tool has been used with options shown below:
# stress -c 65 -i 60 -m 25 --vm-bytes 256M -t 120m -v
-c: number of CPU workers
-i: number of IO workers
-m: number of Memory workers
--vm-bytes: allocate memory size for each memory worker(default 256MB)
-t: the duration of running the test
-v: make the test in verbose mode
With this options, I went through 150 for 5min CPU load and almost 100 for 15min load which is pretty amazing.
It has carried out in a test VM with 2 vCPU and 4GB memory.
# stress -c 65 -i 60 -m 25 --vm-bytes 256M -t 120m -v
top - 17:44:56 up 55 min, 3 users, load average: 151.21, 124.73, 99.78
Tasks: 254 total, 67 running, 187 sleeping, 0 stopped, 0 zombie
Cpu(s): 74.5%us, 24.2%sy, 0.0%ni, 0.0%id, 0.0%wa, 0.0%hi, 1.3%si, 0.0%st
Mem: 3924688k total, 3837772k used, 86916k free, 260k buffers
Swap: 3354616k total, 2968112k used, 386504k free, 4040k cached
PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND
3309 root 20 0 6516 124 104 R 3.2 0.0 0:06.40 stress
3327 root 20 0 6516 124 104 R 3.2 0.0 0:06.40 stress
3330 root 20 0 6516 124 104 R 3.2 0.0 0:06.40 stress
3342 root 20 0 6516 124 104 R 3.2 0.0 0:06.40 stress
3372 root 20 0 6516 124 104 R 3.2 0.0 0:06.41 stress
3385 root 20 0 6516 124 104 R 3.2 0.0 0:06.41 stress
3387 root 20 0 6516 124 104 R 3.2 0.0 0:06.42 stress
3391 root 20 0 6516 124 104 R 3.2 0.0 0:06.41 stress
3393 root 20 0 6516 124 104 R 3.2 0.0 0:06.41 stress
3395 root 20 0 6516 124 104 R 3.2 0.0 0:08.21 stress
3403 root 20 0 6516 124 104 R 3.2 0.0 0:06.41 stress
3405 root 20 0 6516 124 104 R 3.2 0.0 0:06.41 stress
3409 root 20 0 6516 124 104 R 3.2 0.0 0:06.41 stress
3413 root 20 0 6516 124 104 R 3.2 0.0 0:06.41 stress
3417 root 20 0 6516 124 104 R 3.2 0.0 0:06.41 stress
3437 root 20 0 6516 124 104 R 3.2 0.0 0:06.41 stress
3315 root 20 0 6516 124 104 R 2.8 0.0 0:06.40 stress
3324 root 20 0 6516 124 104 R 2.8 0.0 0:06.39 stress
3333 root 20 0 6516 124 104 R 2.8 0.0 0:06.40 stress
3336 root 20 0 6516 124 104 R 2.8 0.0 0:06.39 stress
3339 root 20 0 6516 124 104 R 2.8 0.0 0:06.39 stress
3354 root 20 0 6516 124 104 R 2.8 0.0 0:08.21 stress
3381 root 20 0 6516 124 104 R 2.8 0.0 0:06.40 stress
3389 root 20 0 6516 124 104 R 2.8 0.0 0:06.41 stress
3397 root 20 0 6516 124 104 R 2.8 0.0 0:06.41 stress
3401 root 20 0 6516 124 104 R 2.8 0.0 0:06.40 stress
3407 root 20 0 6516 124 104 R 2.8 0.0 0:06.41 stress
3411 root 20 0 6516 124 104 R 2.8 0.0 0:06.41 stress
3415 root 20 0 6516 124 104 R 2.8 0.0 0:06.41 stress
3429 root 20 0 6516 124 104 R 2.8 0.0 0:06.41 stress
3447 root 20 0 6516 124 104 R 2.8 0.0 0:06.40 stress
3449 root 20 0 6516 124 104 R 2.8 0.0 0:06.40 stress
3453 root 20 0 6516 124 104 R 2.8 0.0 0:06.41 stress
3454 root 20 0 6516 124 104 R 2.8 0.0 0:06.61 stress
39 root 20 0 0 0 0 D 1.6 0.0 0:23.33 kswapd0
3306 root 20 0 6516 124 104 R 1.6 0.0 0:06.74 stress
3312 root 20 0 6516 124 104 R 1.6 0.0 0:06.74 stress
3318 root 20 0 6516 124 104 R 1.6 0.0 0:06.74 stress
3321 root 20 0 6516 124 104 R 1.6 0.0 0:06.74 stress
Following demonestrate the results of dd writing of 1GB to the SAN with different block sizes of 1MB, 10MB and 100MB.
The bigger the block size get, the slower the writing to the SAN becomes:
# dd if=/dev/zero of=/tmp/test22 bs=1M count=1000
1000+0 records in
1000+0 records out
1048576000 bytes (1.0 GB) copied, 77.165 s, 13.6 MB/s
# dd if=/dev/zero of=/tmp/test22 bs=10M count=100
100+0 records in
100+0 records out
1048576000 bytes (1.0 GB) copied, 104.555 s, 10.0 MB/s
# dd if=/dev/zero of=/tmp/test22 bs=100M count=10
10+0 records in
10+0 records out
1048576000 bytes (1.0 GB) copied, 1204.23 s, 871 kB/s
Comments
Post a Comment