Efficiently finding a matching subnet in a very large pool - powershell

Question

I have been writing a script that gathers IPAM info given a provided IP address.
It is working, but my current implementation is highly inefficient.
I am using this script to run checkSubnet, which determines if an IP is in a subnet.
First, I query IPAM to gather this $allSubnets object:
Address CIDR Description VLAN
------- ---- ----------- ----
10.15.10.0 24 DMZ 3000
10.15.11.0 24 Voice 3010
10.15.12.0 24 Wireless 3020
10.15.13.0 28 Management 3030
... ... ... ...
Which is then searched like so:
$testCon = Test-Connection hostname -Count 1
$allSubnets | ForEach-Object {
if((checkSubnet -addr1 ('{0}/{1}' -f $_.Address, $_.CIDR) -addr2 $testCon.IPV4Address.IPAddressToString).Condition -eq $true)
{
[pscustomobject]#{
subnet = ('{0}/{1}' -f $_.Address, $_.CIDR)
desc = $_.Description
}
}
}
This works perfectly for smaller queries.
However, it can take an extremely long time to run against all the items in $allSubnets! Say I want to test 20 IP addresses against a full 2000 subnets, suddenly this query will take a full 2 minutes to complete.
Does anyone have any ideas of how to improve the efficiency of this?

Answer 1

I don't have access to the full list of subnets, but the tests against the ones posted have shown that this is a bit faster:
first a "simpler" function to see if an IP is in a certain range:
Function Find-Subnet ([IPAddress]$SubnetAddress,[byte]$CIDR,[IPAddress]$MatchIP){
[IPAddress]$Mask = [System.Convert]::ToUInt64(('1'*$CIDR).PadRight(32,'0'),2)
return (($SubnetAddress.Address -band $Mask.Address) -eq ($MatchIP.Address -band $Mask.Address))
}
this takes advantage of bit-shifting, the [System.IPAddress] object can be a bit of a performance hit, but it still seems a lot faster than the linked function, and is much more concise, you could always rewrite the [IPAddress] casting as another binary function if you really need the extra performance, as we only use the numerical representation of the Address from it in the end.
then I look to limit the number of searches, when searching against 2000+ subnets this should be a massive performance increase, searching against only the few in the question it was only a slight negative though.
foreach ($Reg in ('(.*\.).*','(.*\.).*\..*','(.*\.).*\..*\..*')){
$Prefix = $ToMatch -replace $Reg,'$1'
Write-Host "Searching subnets beginning with '$($Prefix)'..." -Fore Yellow
$AllSubnets | ? {$_.Address.StartsWith($Prefix)} | ForEach-Object {
if (Find-Subnet -SubnetAddress $_.Address -CIDR $_.CIDR -MatchIP $ToMatch){
$_ ; break
}
}
}
this loops through the regex snippets '(.*\.).*','(.*\.).*\..*' and '(.*\.).*\..*\..*' which when run through the $Prefix = $ToMatch -replace $Reg,'$1' will result in something like this:
'10.11.12.13' -> '(.*\.).*' -> '10.11.12.'
'10.11.12.13' -> '(.*\.).*\..*' -> '10.11.'
'10.11.12.13' -> '(.*\.).*\..*\..*' -> '10.'
we then go through the whole list and pull out subnets with addresses starting with the prefix, for any class A subnet this will be slower than just not including it, for B and C it will be roughly the same performance or faster, with B being more likely to be slower, and C being significantly more likely to be much much quicker.
your current function also does not stop searching after it finds the IP. by including the ;break after returning the matching IP object, we immediately return it and stop searching, if you'd like to store the subnet in a variable you can do $Result = #(foreach ($Reg in ...) { ... }) to end up with it in $Result.