Imagine Delivery Optimization as the office taco run. One brave soul fetches a single order, and suddenly every desk is swimming in salsa without the delivery guy burning through extra mileage.

Missed part 1? We set up the cache server and shipped the Intune policy. Now it is time to watch the magic happen, brag about the bandwidth you saved, and know exactly where to poke when something squeaks.

How the Bits Actually Travel

  • The request A Windows device politely asks Windows Update or Intune for a package.
  • Delivery Optimization (DO) steps in DO checks the Intune policy, notes any Group ID and Cache Host, then calls the DO cloud service for a peer list.
  • Local first
    • Connected Cache – the device tries http://<your-cache> for each block.
    • Peers – at the same time it grabs blocks from friendly neighbors that share the same Group ID.
  • Cloud last Any missing pieces come from Microsoft’s CDN. As soon as a block lands, the device re-shares it so the next client can skip the internet hop.
  • Write-back The cache stores every block it fetched, so the second request is pure LAN traffic.

Bottom line: after the first download, every other PC at that site hums along at LAN speed while your WAN link enjoys a coffee break.

Show Me the Savings

Tenant-Wide Report – WufB Reports Workbook

I’ll cover this at an incredibly high level.

  • Create a Log Analytics Workspace.
  • Visit https://aka.ms/wufbreports.
  • According to the documentation, and in my experience you will not be charged for ingesting WUfB data.
  • Pick your subscription and workspace, then wait. The banner claims 24 hours, yet in reality it often takes 48.

After some time, the workbook will be created. There is a ton of beneficial information in this workbook, but we’re just focused on the DO piece.

Once ready, the workbook shows CDN vs Peer/Cache bytes plus the savings percentage.

We can also see how our content was being distributed across the organization.

My cache server is an Azure VM located in the West US region. Large multi-regional customers love those charts during quarterly reviews.

Per-device snapshotGet-DeliveryOptimizationStatus

  • Real-time counters: BytesFromPeers, BytesFromCacheServer, BytesFromHttp

Quick tip:

If CacheHost is populated but BytesFromCacheServer stays at zero, start with the basics:

  • Can you ping the cache server?
  • Is DNS behaving? (it’s always DNS)
  • Are the required ports and domains open through the firewall?

Monthly roll-up – Get-DeliveryOptimizationPerfSnapThisMonth

Handy for leadership slides – total GB saved this month, and also to make sure the device is functioning as it should.

Continuing on the device itself, you can bring up DO statistics. You should see the majority coming from your cache server or other peers on your network. If you are seeing majority of the traffic coming from Microsoft, then go to the next step in this blog.

When Things Go Sideways

Fallback to CDN?

In Azure, open Connected Caches for Enterprise & Education and check the overview. Is the cache healthy?

  • Give it a reboot – A simple restart often clears things up.
  • Scheduled Tasks check Look at MCC_Monitor_Task. If the Last Run Result shows errors, open the transcript file in your installation folder (for example C:\mccwsl01\WSL_Mcc_Monitor_FromRegisteredTask_Transcript.txt).

The culprit usually reveals itself in the first few lines.

Fine-Tuning Before the Big Rollout – Intune Policies

Cache Host Fallback Delay – These two settings will be really dependent on your environment. This represents the time in seconds before a device abandons the cache and goes to the CDN.

Background Bandwidth Cap – Consider setting business hours and bandwidth percentages, to not take out your network during business hours.

Pilot one branch, tweak the timers, then scale out with confidence.

Key Takeaways

  • Local first, cloud last – understanding the flow makes troubleshooting simple.
  • Measure everything – WUfB reports and PowerShell snapshots prove the ROI fast.
  • Logs are your friend – one glance at BytesFromCacheServer shows whether the cache is earning its keep.
  • Start small, then grow – a tidy pilot keeps surprises to a minimum.