Optimizing lookups in PowerShell

By Anatoly Mironov

Have you had a PowerShell script that contains two bigger arrays and you wanted merge the information. It can become quite slow if you need to search for every item from array A through all items in array B. The solution is called a HashTable! It might be not an advanced tip for some, but I was really glad to see a huge improvement, so I decided to share it as a post.

My Array A ($sites) is a list of SharePoint Sites (over 10K of them). For every site I need to get information on the owner (such as UsageLocation). In order to minimize calls to the server I want to reuse the information - in my array B: $users. This array of users has also thousands of entries.

Here is my main (simplified) setup:

$users = # @() array, code ommitted for brevity
$sites = # @() array, code ommitted for brevity
$sitesAndOwners = $sites | ForEach-Object {
[PSCustomObject]@{
Site = $_
Owner = GetUserInfo($_.Owner)
}
}
view raw optimize-lookup-main.ps1 hosted with ❤ by GitHub

Traversing the array B for the right item for every entry in array A is slow: Where-Object:

function GetUserInfoSlow($upn) {
$user = $users | Where-Object { $_.UserPrincipalName -eq $upn }
if ($user.Count -eq 0) {
$user = Get-AzureADUser -SearchString $upn
$users = $users + $user
}
return $user
}
view raw optimize-lookup-slow.ps1 hosted with ❤ by GitHub

Using a hashtable is much faster:

$sersHash = @{}
function GetUserInfoFast($upn) {
# we check if there is an entry even if value is null
if ($sersHash.Contains($upn)) {
$user = $sersHash[$upn]
}
else {
$user = Get-AzureADUser -SearchString $upn
$sersHash.Add($upn, $user)
}
$user
}
view raw optimize-lookup-fast.ps1 hosted with ❤ by GitHub

In my example it took hours first. Now it takes seconds. A bonus: here is how you can convert an array to a hash table:

#bonus: convert array to a hash table
$users | ForEach-Object {
$usersHash.Add($_.UserPrincipalName, $_)
}
view raw optimize-lookup-convert.ps1 hosted with ❤ by GitHub

That’s all I’ve got today. Sharing is caring… And of course, big thanks to my colleague Anton H. for his advise.