average cpu and memory script

[sureshupadhya]

Hi ,

need to extract cpu and memory usage for 800+ vm's for a day.. the existing once which are there in the community are getting hung , due to which its not completing.

could any one pls help.

thanks

[LucD]

Do you mean the average CPU and Mem for each VM over a 1 day interval.

Or all the metrics for all VMs for 1 day ?

[sureshupadhya]

for each vm 1 cpu usage value which is an average for the day and 1 mem usage value which is an averate for the day.

[LucD]

Try like this

$stats = "cpu.usage.average","mem.usage.average"
$start = (Get-Date).AddDays(-1)
$vms = Get-VM

Get-Stat -Entity $vms -Stat $stats -Start $start |
Group-Object -Property {$_.Entity.Name} | %{
  New-Object PSObject -Property @{
    VM = $_.Group[0].Entity.Name
    Date = $start
    CPUAvg = $_.Group | where {$_.MetricId -eq "cpu.usage.average"} |
      Measure-Object -Property Value -Average |
      Select -ExpandProperty Average
    MemAvg = $_.Group | where {$_.MetricId -eq "mem.usage.average"} |
      Measure-Object -Property Value -Average |
      Select -ExpandProperty Average
  }
}

[LittleNickey]

Luc, I found your old script Discover Memory Overallocations the other day, which I thought was kind of sweet. I guess this is pretty much the same thing. In that function however you get the option to select the time frame.

[Man7630]

Hi,

I was looking for a script which will give me Average CPU and Average Memory statistics (with interval as 30 minutes) for last day. I have somewhere around 500 VMs.

I tried with some of the Scripts from this Community, but those are taking a lot of time which is a main concern for me.

Basically i wanted to schedule a job which will pull CPU and Memory values (with interval as small as 30 min) on a daily basis and store in my local database so that I can do some analysis on such historical data which is collected.

If my jobs runs for a longer durations say (more than 2-3 hours), it will create problem for me.

I want result set in a quick way. Please help.

[LucD]

Could you perhaps share the script you are using ?

An alternative could be to use my Get-Stat2 function, that one tends to be faster than the regular Get-Stat cmdlet.

[Man7630]

Below is the query I used

ForEach ($VM in Get-VM )

{

$todaymidnight = (get-Date -hour 0 -minute 0 -second 0).addminutes(-1)

Get-stat -entity $VM -stat cpu.usage.average -start $todaymidnight.adddays(-1) -finish $todaymidnight | Select Entity, TimeStamp, Value | Export-Csv "C:\vCenter\one.csv" -NoTypeInformation -UseCulture

}

[LucD]

You seem to be calling the Get-Stat cmdlet for each VM, that causes a lot of unnecessary overhead.

You can call Get-Stat for all VM and then use the Group-Object cmdlet to split out the results per VM (see the script I included previously).

[Man7630]

Thanks for the help LucD. I am new to Power Cli and vCenter.

I used below script, but still taking time (running for more than 90 minutes).

In one of my vCenter site, I have 1000 servers, one get-stat command for pulling CPU and Memory for one day 30 min interval taking average 40 seconds, so I believe entire result set will take more than 11 hours, which is not quite acceptable.

Can you suggest a way (is there any other way to pull data from vCenter (APIs or DB) - just guessing...

$stats = "cpu.usage.average","mem.usage.average"

$todaymidnight = (get-Date -hour 0 -minute 0 -second 0).addminutes(-1)

start $todaymidnight.adddays(-1)

get-date

$vms = Get-VM

Get-Stat -Entity $vms -Stat $stats -Start $start |

Group-Object -Property EntityId | Foreach-Object{

  New-Object PSObject -Property @{

    VM = $_.Group[0].Entity.Name

    Date = $_.Group[0].Timestamp.Date

    CPU = $_.Group | Where-Object {$_.MetricId -eq $stats[0]} |

          Measure-Object -Average -Property Value |

          Select -ExpandProperty Average

    Memory = $_.Group | where {$_.MetricId -eq $stats[1]} |

          Measure-Object -Average -Property Value |

          Select -ExpandProperty Average

  }

} | Select VM,Date,CPU,Memory |  Export-Csv "C:\vCenter\two.csv" -NoTypeInformation -UseCulture

[LucD]

I think there might be a typo in the script.

Shouldn't it be like this ?

$stats = "cpu.usage.average","mem.usage.average"

$todaymidnight = (get-Date -hour 0 -minute 0 -second 0).addminutes(-1)

$start = $todaymidnight.adddays(-1)

$finish = $todaymidnight

$vms = Get-VM

Get-Stat -Entity $vms -Stat $stats -Start $start |

Group-Object -Property EntityId | Foreach-Object{

  New-Object PSObject -Property @{

    VM = $_.Group[0].Entity.Name

    Date = $_.Group[0].Timestamp.Date

    CPU = $_.Group | Where-Object {$_.MetricId -eq $stats[0]} |

          Measure-Object -Average -Property Value |

          Select -ExpandProperty Average

    Memory = $_.Group | where {$_.MetricId -eq $stats[1]} |

          Measure-Object -Average -Property Value |

          Select -ExpandProperty Average

  }

} | Select VM,Date,CPU,Memory |  Export-Csv "C:\vCenter\two.csv" -NoTypeInformation -UseCulture

That script is calculating the averages of the complete previous day for CPU and memory for each VM.

Is that what you are looking for ?

You should time this with more than 1 VM on the Entity parameter.

The advantage is that by passing multiple entities, you avoid a lot of the repetitive overhead that an individual Get-Stat call has.

[Man7630]

Sorry I was looking for granular data (like 30 minutes interval) for the last day ?

Can you provide correct script for that ?