ESXi Host Capacity Performance Report

[DaddyDee10]

Greetings all,

I'm presently in the process of knocking up a PowerCLI script to report on all ESXi hosts across vCenters instances (x3).

I came across the following group discussion on here which assisted immensely (see below link:)

powercli script to capture cpu & mem usage stats

I've since tweaked script ensuring just Hosts are to be reported:

Connect-VIServer <server> -User <user> -Password <password>

$allhosts = @()
$hosts = Get-VMHost

foreach($vmHost in $hosts){
  $hoststat = "" | Select HostName, MemMax, MemAvg, MemMin, CPUMax, CPUAvg, CPUMin
  $hoststat.HostName = $vmHost.name
 
  $statcpu = Get-Stat -Entity ($vmHost)-start (get-date).AddDays(-30) -Finish (Get-Date)-MaxSamples 10000 -stat cpu.usage.average
  $statmem = Get-Stat -Entity ($vmHost)-start (get-date).AddDays(-30) -Finish (Get-Date)-MaxSamples 10000 -stat mem.usage.average

  $cpu = $statcpu | Measure-Object -Property value -Average -Maximum -Minimum
  $mem = $statmem | Measure-Object -Property value -Average -Maximum -Minimum
 
  $hoststat.CPUMax = $cpu.Maximum
  $hoststat.CPUAvg = $cpu.Average
  $hoststat.CPUMin = $cpu.Minimum
  $hoststat.MemMax = $mem.Maximum
  $hoststat.MemAvg = $mem.Average
  $hoststat.MemMin = $mem.Minimum
  $allhosts += $hoststat
}
$allhosts | Select HostName, MemMax, MemAvg, MemMin, CPUMax, CPUAvg, CPUMin | Export-Csv "c:\Hosts.csv" -noTypeInformation

In addition, both Network I/O and disk I/O (Datastore specific) would need to be captured too.

Your ideas & assistance would be greatly appreciated.

Thanks in advance.

[LucD]

You can do something like this.

But note that the disk statistics are the aggregate, in other words the average over all datastores.

If you need this per datastore, you will need to use one or more of the datastore metrics (read and/or write).

If you can have varying numbers of datastore on each host, how do you intend to present this ?

Connect-VIServer <server> -User <user> -Password <password>

$allhosts = @()

$hosts = Get-VMHost

$start = (Get-Date).AddDays(-30)

$stat = 'cpu.usage.average','mem.usage.average','net.usage.average','disk.usage.average'

Get-Stat -Entity $hosts -Stat $stat -Start $start |

Group-Object -Property {$_.Entity.Name} | %{

  $cpuStat = $_.Group | where{$_.MetricId -eq 'cpu.usage.average'} |

    Measure-Object -Property value -Average -Maximum -Minimum

  $memStat = $_.Group | where{$_.MetricId -eq 'mem.usage.average'} |

    Measure-Object -Property value -Average -Maximum -Minimum

  $netStat = $_.Group | where{$_.MetricId -eq 'net.usage.average' -and $_.Instance -eq ''} |

    Measure-Object -Property value -Average -Maximum -Minimum

  $diskStat = $_.Group | where{$_.MetricId -eq 'disk.usage.average' -and $_.Instance -eq ''} |

    Measure-Object -Property value -Average -Maximum -Minimum

       

  [pscustomobject]@{

    HostName = $_.Group[0].Entity.Name

    CPUAvg = $cpuStat.Average

    CPUMin = $cpuStat.Minimum

    CPUMax = $cpuStat.Maximum

    MemAvg = $cpuStat.Average

    MemMin = $cpuStat.Minimum

    MemMax = $cpuStat.Maximum

    NetAvg = $cpuStat.Average

    NetMin = $cpuStat.Minimum

    NetMax = $cpuStat.Maximum

    DiskAvg = $cpuStat.Average

    DiskMin = $cpuStat.Minimum

    DiskMax = $cpuStat.Maximum

  }

}

[DaddyDee10]

I really appreciate your prompt response LucD.

I'm pretty much new to PowerCLI and relying heavily on feedback from the community on this.

In answer to your question?  I guess what would be the easiest approach?  As of yet, there is no reporting mechanism in place on ESXi performance within team.  The following attributes need to be reported on:  active CPUs, CPU usage, Mem usage, Network I/O, Disk I/O.

So let's assume disk is local to ESXi host?  How would this be reflected in final PowerCLI script?

Any help would be greatly appreciated.

[LucD]

Can you clarify what you want to see under "active CPUs" ?

If I understand correctly there is only a local datastore connected to the ESXi, correct ?

Which metric do you want to see for "Disk I/O" ?

Have a look under the Performance tab in the client and determine which metric you are interested in, and then let me know

[DaddyDee10]

I'm waiting on feedback from colleague.

I'll reply as soon as I know.

Many thanks for all your help.:smileyhappy:

[DaddyDee10]

Hi LucD,

Metrics for CPU counter, should remain as they are (e.g. average, min and max).

Datastore metrics should be:

Read latency

Write latency

Network metrics should be:

Data transmit rate

Data receive rate

Usage

Finalized script will be run against all vCenter instances within environment.

In addition, my colleague has asked that raw informaton be extracted from the daily VC DB pertaining to ESXi Host CPU usage during the day!!  He has asked this due to above script only capuring the averages over the last (specified) time period and is concerned daily info may get lost.  I'm guessing, you'd suggest tweaking specified time period in script to capture this on a day to day basis, hence an additional script being run solely for ESXi Host usage?

Any help would be kindly appreciated.

[LucD]

This should give you the request counters, but note that some of the metrics might require a specific Statistics level.

My Stats Toolbox – A vSphere Server Performance Counter tool post provides a tool that will allow you to check that.

Connect-VIServer <server> -User <user> -Password <password>

$allhosts = @()

$hosts = Get-VMHost

$start = (Get-Date).AddDays(-30)

$stat = 'cpu.usage.average','mem.usage.average','net.usage.average','net.bytesRx.average','net.bytesTx.average',

    'datastore.totalReadLatency.average','datastore.totalWriteLatency.average'

Get-Stat -Entity $hosts -Stat $stat -Start $start |

Group-Object -Property {$_.Entity.Name} | %{

  $cpuStat = $_.Group | where{$_.MetricId -eq 'cpu.usage.average'} |

    Measure-Object -Property value -Average -Maximum -Minimum

  $memStat = $_.Group | where{$_.MetricId -eq 'mem.usage.average'} |

    Measure-Object -Property value -Average -Maximum -Minimum

      

  [pscustomobject]@{

    HostName = $_.Group[0].Entity.Name

    CPUAvg = $cpuStat.Average

    CPUMin = $cpuStat.Minimum

    CPUMax = $cpuStat.Maximum

    MemAvg = $cpuStat.Average

    MemMin = $cpuStat.Minimum

    MemMax = $cpuStat.Maximum

    NetUsage = $_.Group | where{$_.MetricId -eq 'net.usage.average'} | Measure-Object -Property Value -Average | Select -ExpandProperty Average

    NetTr = $_.Group | where{$_.MetricId -eq 'net.bytesTx.average'} | Measure-Object -Property Value -Average | Select -ExpandProperty Average

    NetRx = $_.Group | where{$_.MetricId -eq 'net.bytesRx.average'} | Measure-Object -Property Value -Average | Select -ExpandProperty Average

    DSReadLatency = $_.Group | where{$_.MetricId -eq 'datastore.totalReadLatency.average'} | Measure-Object -Property Value -Average | Select -ExpandProperty Average

    DSWriteLatency = $_.Group | where{$_.MetricId -eq 'datastore.totalWriteLatency.average'} | Measure-Object -Property Value -Average | Select -ExpandProperty Average

  }

}

For your other question, that should indeed be another daily task.

[DaddyDee10]

You're a life saver!! :smileyhappy:

Many thanks for all your help and will refer to your Stats Toolbox.

[DaddyDee10]

Hi LucD,

Due to the enormity of our environment (datacenters, clusters, etc), I've decided to run script per vCenter instance as opposed to multiple instances all at once.

I've noticed after perfoming a cut and paste of script (see attached), cursor just remains idle and blinks after the '-noTypeInformation.'

Is this to be expected?

This again occurred on Friday and I allowed script to run for a few hours and nothing happened!!

Any assistance would be welcome?

[LucD]

I would suggest you place the updated script in a .ps1 file, and then run that .ps1 file from the PowerCLI prompt (instead of pasting the script to the command prompt).

Have a look at XtraVirt's Beginner's Guide.

[DaddyDee10]

Many thanks.. :smileywink: