Marka Mekapse
asked on
Slow performance on Windows 2016 Server
Recently migrated SQL database from 2008 to 2016 (SQL 2008 to SQL 2014); the issue appears to be on the Operating System as i have run several tests that prove this. I run a simple script that runs a for loop counting from 1 to 16000 to emulate a rudimentary write to disk. When i run this script on a 2016 server it takes almost 6 minutes to complete where as it is instantaneous on a windows 2008 server.
The purpose of the test was to prove that there is additional overhead on a Server 2016 box versus the 2008.
1. 2008 server
a. On the <host> network
i. Result = instant
b. Off the <host> network
i. Result = instant
2. 2012 server
a. On the <host> network
i. Result = 35 seconds
b. Off the <host> network
i. Result = Test 1 - 15 seconds (SSD) ;
ii. Result = Test 2 - 29 Seconds
3. 2016 server
a. On the <host> network
i. Result = 5 minutes
b. Off the <host> network
i. Result = 6 minutes
We know it is not the disk or the amount of memory and CPU
Is there anything i can do to the windows server 2016 box to increase its performance and remove the "overhead"???
Server OS is Server 2016 - 4 vCPU w 2 Cores & 50 GB Ram
Virtualization - Vmware 5.5 U2
SQL server 2014
here is my script that i have been testing with
The purpose of the test was to prove that there is additional overhead on a Server 2016 box versus the 2008.
1. 2008 server
a. On the <host> network
i. Result = instant
b. Off the <host> network
i. Result = instant
2. 2012 server
a. On the <host> network
i. Result = 35 seconds
b. Off the <host> network
i. Result = Test 1 - 15 seconds (SSD) ;
ii. Result = Test 2 - 29 Seconds
3. 2016 server
a. On the <host> network
i. Result = 5 minutes
b. Off the <host> network
i. Result = 6 minutes
We know it is not the disk or the amount of memory and CPU
Is there anything i can do to the windows server 2016 box to increase its performance and remove the "overhead"???
Server OS is Server 2016 - 4 vCPU w 2 Cores & 50 GB Ram
Virtualization - Vmware 5.5 U2
SQL server 2014
here is my script that i have been testing with
#Powershell Script#
#
$file = "C:\testlog.txt"
Remove-Item –path $file
$d = Get-Date -Format G
Write-Output $d >> $file
for($i=1
$i -le 16000
$i++){
Write-Output $i >> $file
}
$d = Get-Date -Format G
Write-Output $d >> $file
#End
ASKER CERTIFIED SOLUTION
membership
Create a free account to see this answer
Signing up is free and takes 30 seconds. No credit card required.
I also found a different script that performs an IO benchmark (Original is here but it is buggy as crap, so I fixed some bugs and attached the debugged version)
This is a more realistic drive IO test that does things at a more fundamental level. If you run this script against both systems, you should see much more realistic speeds and that they're very close (unless you have significantly different Disk architecture).
The difference we're seeing with using the >> operation is likely due to increased complexity and more features in Powershell on 2016 than other version. I don't know exactly what the cause is, but it is a matter of how we're interacting with the files.
disktest.ps1
This is a more realistic drive IO test that does things at a more fundamental level. If you run this script against both systems, you should see much more realistic speeds and that they're very close (unless you have significantly different Disk architecture).
The difference we're seeing with using the >> operation is likely due to increased complexity and more features in Powershell on 2016 than other version. I don't know exactly what the cause is, but it is a matter of how we're interacting with the files.
disktest.ps1
ASKER
Thank You