Network drive inaccessible or slow over VPN: find the cause

Troubleshoot slow or inaccessible VPN network drives: DNS, routes, TCP 445, permissions, DFS and SMB performance with commands and pilot validation.

01

A connected VPN does not guarantee file access

A slow network drive over VPN can originate on the endpoint, tunnel, DNS, routing, permissions or server. Record the time, exact error, UNC path and affected users. Compare the same account and file in the office and remotely. A missing drive letter and an inaccessible share are different symptoms.

Examples use fictional fichiers.atelier.example and 10.20.30.40. Replace them with approved targets. This guide supports focused troubleshooting rather than a complete network assessment.

02

1. Establish scope and prerequisites

Use Windows PowerShell with built-in Windows modules on a supported Windows release. Record VPN client and gateway versions and check vendor compatibility and licensing. Get-VpnConnection may not describe profiles managed by Cisco, Fortinet or other third-party clients.

Start in the affected user session. An elevated console can have a different mapped-drive context. Some logs or policy reads need additional permissions: request only the necessary access, not Domain Admin for local collection.

PowerShell — read-only collection
# Local inventory; does not change configuration
Get-CimInstance Win32_OperatingSystem | Select-Object Caption,Version,BuildNumber
Get-SmbMapping | Select-Object LocalPath,RemotePath,Status
Get-NetIPConfiguration
Get-VpnConnection | Select-Object Name,ConnectionStatus,SplitTunneling
03

2. Check the name, DNS and suffixes

Compare the short name with the FQDN. Incorrect DNS answers, missing suffixes or unsuitable NRPT rules can direct requests incorrectly. Record A and AAAA answers; correct resolution does not prove reachability. Do not replace corporate DNS with a public resolver as a workaround.

PowerShell — read-only collection
# Read the effective client configuration
Get-DnsClientServerAddress
Get-DnsClientGlobalSetting
Get-DnsClientNrptPolicy -Effective
Resolve-DnsName fichiers.atelier.example
Resolve-DnsName fichiers
04

3. Check the actual route to each destination

An active tunnel may lack routes to file services. Overlapping home and corporate subnets can also misdirect traffic. Examine the selected interface and source address, then have the administrator inspect the return path and gateway policies. Repeat for IPv6 when an AAAA address is used. Full-tunnel mode does not eliminate the need to inspect more specific routes.

PowerShell — read-only collection
# Replace with the approved server address
Find-NetRoute -RemoteIPAddress 10.20.30.40
Get-NetRoute | Select-Object DestinationPrefix,NextHop,InterfaceAlias,RouteMetric
05

4. Separate TCP, SMB and authorization

For a conventional SMB share, test TCP 445. Success confirms a TCP connection, not share access or permissions. Failure can involve routing, filtering or the target service; it does not identify the VPN as the sole cause. Never expose SMB 445 directly to the Internet to repair access.

After an authorized share-access attempt, inspect SMB connections. An empty list may only mean no session exists. Preserve signing and encryption; disabling them is not a generic fix.

PowerShell — read-only collection
# Connectivity probe; no configuration change
Test-NetConnection fichiers.atelier.example -Port 445 -InformationLevel Detailed
Get-SmbConnection | Select-Object ServerName,ShareName,UserName,Dialect,Encrypted
06

5. Check identity, permissions, DFS and mapping

Open the approved UNC path in the affected session. If it works but the drive letter does not, inspect the mapping and whether its GPO or script runs before the VPN connects. With DFS, identify the actual target: the namespace may work while its target is unreachable.

Inspect groups and tickets without purging them. IP-based access can change authentication and is not sufficient proof of a DNS fault. Have the owner review share and NTFS permissions; do not grant Everyone Full Control as a test.

PowerShell — read-only collection
# Current-session identity and tickets; sensitive output
whoami /groups
klist
07

6. Measure slowness with a repeatable scenario

Compare a large file with a small set of fictional files in an approved test location. Copying writes data: treat it as a controlled test, plan space and cleanup, avoid production folders and mirror or delete options. Record duration, volume, time, wired or Wi-Fi access and gateway load.

Small files can magnify latency effects. High Internet throughput does not guarantee SMB performance. Compare with the LAN and monitor storage, server CPU and VPN CPU. Do not randomly change MTU, compression or security; establish packet loss, retransmissions and hardware constraints before a pilot.

08

7. Collect bounded, correlated evidence

List available channels first. The following read covers two hours and at most 100 events. No events does not rule out a fault: the channel may be disabled or retention insufficient. Correlate timestamps and time zones with the server and gateway.

If packet capture is needed, agree on a short duration, interfaces and targets with the administrator. Captures and exports create files and may disclose names, addresses and sensitive data. Restrict access, define deletion and avoid uploading raw captures to public tickets.

PowerShell — read-only collection
# Discover available logs before querying
Get-WinEvent -ListLog 'Microsoft-Windows-SMBClient/*' |
  Select-Object LogName,IsEnabled,RecordCount,MaximumSizeInBytes
Get-WinEvent -FilterHashtable @{
  LogName='Microsoft-Windows-SMBClient/Connectivity'
  StartTime=(Get-Date).AddHours(-2)
} -MaxEvents 100 | Select-Object TimeCreated,Id,LevelDisplayName,Message
09

8. Pilot the correction and validate rollback

Every DNS, route, VPN profile, ACL or GPO change needs a hypothesis, owner and saved prior configuration. Test one change on a small group in an authorized window. Validate opening, reading, approved writing, reconnection and denial for an unauthorized account. Compare the same before-and-after measurements.

Define stop criteria and an exact restoration procedure: previous route or profile, scope and owner. Do not remove every SMB connection or globally disable the firewall. A useful deliverable includes the timeline, test results, confirmed cause, correction and business validation. Infrastructure consulting can support the investigation and pilot.

10

Microsoft sources