PowerShell inner-join large Excel table is extremely slow

Perhaps using this will be faster:

# create a Hashtable, which works case-insensitive by default
$right = @{}
foreach ($name in $df2.Name) { $right[$name] = $true }  # the value doesn't matter

# now loop over the left array ($df1) and output only 
# the rows where the name is found in the $right Hashtable
$result = foreach ($item in $df1) {
    if ($right.ContainsKey($item.Name)) {
        $item
    }
}

If you want to compare the names case-sensitively, you can use a HashSet:

$right = [System.Collections.Generic.HashSet[String]]::new()
foreach ($name in $df2.Name) { [void]$right.Add($name) } 

# now loop over the left array ($df1) and output only 
# the rows where the name is found in the $right HashSet
$result = foreach ($item in $df1) {
    if ($right.Contains($item.Name)) {
        $item
    }
}

You can also try if Compare-Object would be faster in your case:

$result = Compare-Object -ReferenceObject $df1 -DifferenceObject $df2 -Property Name -IncludeEqual -ExcludeDifferent -PassThru

^{Compare-Object also has a -CaseSensitive switch if that is what you need}

Based on the comment you posted, you can use the first option mentioned above and change the final loop into this:

# now loop over the left array ($df1) and output only 
# the rows where the name is found in the $right Hashtable
# an inner loop merges the properties 
$result = foreach ($item in $df1) {
    if ($right.ContainsKey($item.Name)) {
        $merge = $df2 | Where-Object {$_.Name -eq $item.Name}
        $merge.PsObject.Properties.Name | 
            Where-Object {$item.PsObject.Properties.Name -notcontains $_} | 
            ForEach-Object {
                $item | Add-Member -MemberType NoteProperty -Name $_ -Value $merge."$_"
            }
        $item
    }
}

$result

The .NET method Linq.Enumerable.Join seems to be pretty fast. As it's not immediately obvious how to use it from PowerShell (you need some casts to be able to pass ScriptBlocks to its selector parameters), I've wrapped it in a PowerShell function Join-Enumerable to make it easier to use.

Function Join-Enumerable{
    [CmdletBinding()]
    param (
        [Parameter(Mandatory)] 
        [Collections.IEnumerable] $Left,
        
        [Parameter(Mandatory)] 
        [Collections.IEnumerable] $Right,

        [Parameter(Mandatory)] 
        $LeftKey,
        
        [Parameter(Mandatory)] 
        $RightKey,      
        
        [Parameter(Mandatory)] 
        [scriptblock] $Result,

        [Parameter()] 
        [switch] $NoEnumerate
     )

    if( $RightKey -isnot [scriptblock] ) {
        $RightKey = "$RightKey" -replace "'", "''"
        $RightKey = [scriptblock]::Create( "`$args[0].'$RightKey'" )
    }
    if( $LeftKey -isnot [scriptblock] ) {
        $LeftKey = "$LeftKey" -replace "'", "''"
        $LeftKey = [scriptblock]::Create( "`$args[0].'$LeftKey'" )
    }

    $enumerator = [Linq.Enumerable]::Join( $Left, $Right, 
        [Func[object, object]] $RightKey, 
        [Func[object, object]] $LeftKey, 
        [Func[object, object, object]] $Result )

    if( $NoEnumerate ) {
        return , $enumerator  # Outputs the enumerator itself
    }

    $enumerator  # Enumerates and outputs each result element
}

Demo:

$df1 = ConvertFrom-Csv @'
Name,Foo
foo,1
bar,2
baz,3
'@

$df2 = ConvertFrom-Csv @'
Name,Bar
bar,4
bam,5
foo,6
'@

$joinedArray = Join-Enumerable -Left $df1 -Right $df2 -LeftKey Name -RightKey Name -Result {
    [PSCustomObject] @{ 
        Name = $args[0].Name
        Foo  = $args[0].Foo
        Bar  = $args[1].Bar
    }
}

$joinedArray  # Output

Output:

Name Foo Bar
---- --- ---
foo  1   6
bar  2   4

Remarks:

• Parameters -Left and -Right specify the two arrays (more exactly anything that implements the IEnumerable interface) to join
• Parameters -LeftKey and -RightKey specify the keys to join on from elements of the -Left and -Right arguments. These can be either the name of a property or a scriptblock that gets the current element passed as an argument and outputs the key value.
• Parameter -Result takes a scriptblock that specifies how to generate each output object. It takes two arguments, the 1st is an element from the -Left argument and the 2nd is an element from the -Right argument.
• While Linq.Enumerable.Join doesn't actually return the joined objects, but an enumerator that lazily evaluates the result, PowerShell commands are expected to output the actual objects. This is what the function does by default. If you really need the enumerator, pass -NoEnumerate to the function.
• For maximum performance, arrays of class objects should be passed for -Left and -Right and the result should be instanciated using a class instead of PSCustomObject. If you don't control the type of the input objects, using a class for the output objects still gives a noticable performance improvement and uses less memory:

  class FooBar { 
      [string] $Name
      [int] $Foo
      [int] $Bar
  }
  $joinedArray = Join-Enumerable -Left $df1 -Right $df2 -LeftKey Name -RightKey Name -Result {
      [FooBar] @{ 
          Name = $args[0].Name
          Foo  = $args[0].Foo
          Bar  = $args[1].Bar
      }
  }
  

Performance tests:

For comparing performance I used the this code. Each test script creates two arrays of 100000 elements each, with random name properties. The name properties of the 2nd array match elements from the 1st array randomly. 10 runs are done for each method and the average duration is taken.