Difference between revisions of "Brytlyt Testing"
(Created page with "== Installation == Nvidia-docker will be needed to run our GPU software, link with installation instructions: https://github.com/NVIDIA/nvidia-docker/tree/1.0 I prepared 2 i...") |
|||
| (One intermediate revision by the same user not shown) | |||
| Line 3: | Line 3: | ||
Nvidia-docker will be needed to run our GPU software, link with installation instructions: https://github.com/NVIDIA/nvidia-docker/tree/1.0 | Nvidia-docker will be needed to run our GPU software, link with installation instructions: https://github.com/NVIDIA/nvidia-docker/tree/1.0 | ||
| − | I prepared 2 images, one with BrytlytDB cluster set for 1 gpu and another one with cluster for 2 GPU's | + | I prepared 2 images, one with BrytlytDB cluster set for 1 gpu and another one with cluster for 2 GPU's. '''Note that you need a docker account with read access to the brytlyt repo to pull these images''' |
<syntaxhighlight> | <syntaxhighlight> | ||
nvidia-docker pull brytlytdb/test:1xp100 | nvidia-docker pull brytlytdb/test:1xp100 | ||
| Line 26: | Line 26: | ||
== Testing == | == Testing == | ||
| + | |||
| + | === Sysbench testing === | ||
| + | |||
| + | <syntaxhighlight> | ||
| + | [root@brytlyt-client ~]# yum install epel-release -y | ||
| + | [root@brytlyt-client ~]# yum install sysbench | ||
| + | [root@brytlyt-client ~]# sysbench --table-size=1000000 --pgsql-host=192.168.0.4 --pgsql-port=5432 --pgsql-user=mouza --pgsql-db=mouza_test --db-driver=pgsql /usr/share/sysbench/oltp_read_only.lua prepare | ||
| + | [root@brytlyt-client ~]# sysbench --table-size=1000000 --pgsql-host=192.168.0.4 --pgsql-port=5432 --pgsql-user=mouza --pgsql-db=mouza_test --db-driver=pgsql --num-threads=4 --max-requests=100000 /usr/share/sysbench/oltp_read_only.lua run | ||
| + | WARNING: --num-threads is deprecated, use --threads instead | ||
| + | WARNING: --max-requests is deprecated, use --events instead | ||
| + | sysbench 1.0.9 (using system LuaJIT 2.0.4) | ||
| + | |||
| + | Running the test with following options: | ||
| + | Number of threads: 4 | ||
| + | Initializing random number generator from current time | ||
| + | |||
| + | |||
| + | Initializing worker threads... | ||
| + | |||
| + | Threads started! | ||
| + | |||
| + | SQL statistics: | ||
| + | queries performed: | ||
| + | read: 70952 | ||
| + | write: 0 | ||
| + | other: 10136 | ||
| + | total: 81088 | ||
| + | transactions: 5068 (506.48 per sec.) | ||
| + | queries: 81088 (8103.72 per sec.) | ||
| + | ignored errors: 0 (0.00 per sec.) | ||
| + | reconnects: 0 (0.00 per sec.) | ||
| + | |||
| + | General statistics: | ||
| + | total time: 10.0040s | ||
| + | total number of events: 5068 | ||
| + | |||
| + | Latency (ms): | ||
| + | min: 5.23 | ||
| + | avg: 7.89 | ||
| + | max: 19.58 | ||
| + | 95th percentile: 10.27 | ||
| + | sum: 39996.59 | ||
| + | |||
| + | Threads fairness: | ||
| + | events (avg/stddev): 1267.0000/16.67 | ||
| + | execution time (avg/stddev): 9.9991/0.00 | ||
| + | </syntaxhighlight> | ||
Latest revision as of 11:51, 26 February 2018
Installation
Nvidia-docker will be needed to run our GPU software, link with installation instructions: https://github.com/NVIDIA/nvidia-docker/tree/1.0
I prepared 2 images, one with BrytlytDB cluster set for 1 gpu and another one with cluster for 2 GPU's. Note that you need a docker account with read access to the brytlyt repo to pull these images
nvidia-docker pull brytlytdb/test:1xp100
#and respectively:
nvidia-docker pull brytlytdb/test:2xp100To launch BrytlytDB in the background run this 2 commands:
nvidia-docker run -di -p 5432:5432 --name=brytlyt brytlytdb/test:1xp100
nvidia-docker exec -di brytlyt /root/brytlyt startThey will start the container with BrytlytDB cluster in the background with port 5432 open, so that it can be accessed by any Postgres interface. To launch Postgres terminal from the container use:
nvidia-docker exec -ti brytlyt /usr/local/brytlyt/bin/psql brytlyt brytlytHere is a link to our database documentation and instruction on how to create tables that utilises GPU's, https://www.brytlyt.com/documentation/data-definition-ddl/tables/
Testing
Sysbench testing
[root@brytlyt-client ~]# yum install epel-release -y
[root@brytlyt-client ~]# yum install sysbench
[root@brytlyt-client ~]# sysbench --table-size=1000000 --pgsql-host=192.168.0.4 --pgsql-port=5432 --pgsql-user=mouza --pgsql-db=mouza_test --db-driver=pgsql /usr/share/sysbench/oltp_read_only.lua prepare
[root@brytlyt-client ~]# sysbench --table-size=1000000 --pgsql-host=192.168.0.4 --pgsql-port=5432 --pgsql-user=mouza --pgsql-db=mouza_test --db-driver=pgsql --num-threads=4 --max-requests=100000 /usr/share/sysbench/oltp_read_only.lua run
WARNING: --num-threads is deprecated, use --threads instead
WARNING: --max-requests is deprecated, use --events instead
sysbench 1.0.9 (using system LuaJIT 2.0.4)
Running the test with following options:
Number of threads: 4
Initializing random number generator from current time
Initializing worker threads...
Threads started!
SQL statistics:
queries performed:
read: 70952
write: 0
other: 10136
total: 81088
transactions: 5068 (506.48 per sec.)
queries: 81088 (8103.72 per sec.)
ignored errors: 0 (0.00 per sec.)
reconnects: 0 (0.00 per sec.)
General statistics:
total time: 10.0040s
total number of events: 5068
Latency (ms):
min: 5.23
avg: 7.89
max: 19.58
95th percentile: 10.27
sum: 39996.59
Threads fairness:
events (avg/stddev): 1267.0000/16.67
execution time (avg/stddev): 9.9991/0.00