Vdbench Multi-node test file system read/write

Vdbench Multi-node test file system read/write

Operation Steps:

1. First install Java on each node that needs to be tested (including the master node)

2. then ensure that the hostname of each node is different, and it is best to shut down each firewall

3. mount the networked storage you want to test on each node

4. then operate each node sequentially, enter the Vdbench directory with cmd and run "vdbenchrsh" (Enter the listening mode)

5. Open The host file to the master, and write the IP and hostname of each node .

6. to the master node, use cmd to enter the Vdbench directory, run "vdbench–f parmfile" (parmfile is a pre-configured configuration file)

Write operation configuration file :

Hd=default,vdbench=c:\vdbench,shell=vdbench

hd=hd1,system=10.20.16.96

hd=hd2,system=10.20.16.98

hd=hd3,system=10.20.16.99

fsd=default,depth=1,width=1,files=3,size=5g

Fsd=fsd1,anchor=z:\\1

Fsd=fsd2,anchor=z:\\2

Fsd=fsd3,anchor=z:\\3

Fwd=default,operation=write,xfersize=1m,threads=2,fileio=random,fileselect=random

Fwd=fwd1,fsd=fsd1,host=hd1

Fwd=fwd2,fsd=fsd2,host=hd2

Fwd=fwd3,fsd=fsd3,host=hd3

Rd=rd1,fwd= (FWD1-FWD3), Fwdrate=max,format=yes

Read operation configuration file:

Hd=default,vdbench=c:\vdbench,shell=vdbench

hd=hd1,system=10.20.16.96

hd=hd2,system=10.20.16.98

hd=hd3,system=10.20.16.99

fsd=default,depth=1,width=1,files=3,size=5g

Fsd=fsd1,anchor=z:\\1

Fsd=fsd2,anchor=z:\\2

Fsd=fsd3,anchor=z:\\3

Fwd=default,operation=read,xfersize=1m,threads=2,fileio=random,fileselect=random

Fwd=fwd1,fsd=fsd1,host=hd1

Fwd=fwd2,fsd=fsd2,host=hd2

Fwd=fwd3,fsd=fsd3,host=hd3

Rd=rd1,fwd= (FWD1-FWD3), Fwdrate=max,format=no

Configuration File parameter interpretation:

is not created

hd	Host definition (hd=default is the default configuration content that will be applied to every HD you define )
system	ip address or network name
FSD	name of the file system definition
depth	number of levels to create under locators
width	number of directories to create under locators
files	number of files to create at the lowest level
sizes	file size to be created
anchor
Fwd	Name of the file system workload definition
Operation	mkdir,rmdir,Create,Delete,Open,Close,Read,Write,GetAttrand theSetAttr. Select the individual file action to perform.
Xfersizes	Data block size handled by data transfer (read and write operations)
threads	1 files (the number of threads cannot be greater than the number of files)
fileio	random or sequential , which represents the file I/o The way it will be executed
fileselect	random or sequential to identify how to select a file or directory
rd	run definition
fwdrate
format	yes/no/only/restart/clean/ directories yes : Delete the previous file structure, recreate the directories and files according to the configuration file no : The original file directory structure is used by default and the file
elapsed	30s
Interval	Report interval ordinal, default 1s

Report Interpretation:

The default output report is in the output folder under the program directory

(You can also specify the report location "Vdbench–f parmfile–o C:\test\" When you run the program)

in the default report folder, mainly see summary.html and totals.html

summary.html: Record all the data information

totals.html: Record the average data after all data has been computed

Attention:

1. test Environment: Multi-node,Windows, file system, for linux/ block device test please check the manual.

2. configuration file name customization, more configuration file parameters, please check the manual.

Appendix: ( single node sample configuration file as follows )

fsd=fsd1,anchor=c:\\test,depth=1,width=1,files=2,size=5g

Fwd=fwd1,fsd=fsd1,operation=read,xfersize=1m,threads=1,fileio=random,fileselect=random

Rd=rd1,fwd=fwd1,fwdrate=max,format=no,elapsed=30

Vdbench Multi-node test file system read/write