Identifying which specific process is responsible for a CPU spike requires per-process CPU utilization data, not just an aggregate view of total processor load, so the Data Collector Set needs both a system-wide baseline counter and a per-process counter capable of ranking individual processes against each other by actual CPU consumption. Processor Information(_Total)\% Processor Time supplies the overall CPU utilization context for the whole VM during the spike window, confirming when the spike occurs and how severe it is, while Process(*)\% Processor Time, collected for every process instance, reports each individual process ' s own percentage of CPU time over the same interval, which is exactly the metric needed to identify the single process consuming the most CPU when the spike hits. Process(*)\Thread count only reports how many threads a process currently has, which does not correlate directly with CPU consumption, since a process can hold many idle threads or few very busy ones, so it cannot be used to rank processes by actual CPU usage. The third option pairs a per-process counter incorrectly scoped to the _Total instance (which does not exist as a discrete process to report on in the same way) with the object-level Processor Time counter rather than the per-instance Process object, making it unsuitable for isolating an individual offending process.
Only the combination of the total processor counter and the per-process % Processor Time counter provides the data needed to pinpoint the responsible process.

When any Azure VM extension, including Azure Performance Diagnostics, fails to install, Microsoft ' s documented first troubleshooting step is to sign in to the virtual machine and inspect the WaAppAgent.log file under C:\WindowsAzure\Logs, which records every extension enable, install, and start action the Azure guest agent attempted, along with the specific error code and message it encountered for each attempt. Because installing an extension is a control-plane operation carried out through Azure Resource Manager, it is automatically recorded, with its success or failure status and error detail, in the virtual machine ' s Activity log in the Azure portal with no prior configuration required, making that log the natural second place to check for the same information without needing to sign in to the VM at all. The Alerts blade only surfaces events for alert rules that were explicitly created ahead of time, so it would show nothing about this extension failure unless such a rule already existed and happened to cover it, and the MonitoringAgentlog file relates to the older, now-retired Log Analytics or Microsoft Monitoring Agent rather than to the Performance Diagnostics extension ' s own installation process. Therefore, verifying WaAppAgent.log on VM1 and viewing the Activity log in the Azure portal are the two correct ways to identify the cause.

Explanation:
When a Storage Spaces Direct pool contains three distinct drive tiers, in this case SSD, NVMe, and HDD as shown in the exhibit, Storage Spaces Direct automatically designates only the single fastest media type present, NVMe, as the cache devices for the pool, leaving both the SSD and HDD drives entirely to provide usable pool capacity rather than splitting cache duty across more than one tier. The default cache behavior that Storage Spaces Direct applies then differs depending on which capacity tier is being cached: cached HDD capacity drives receive full read-and-write caching, because HDDs benefit the most from having both random read and random write operations absorbed and accelerated by the much faster flash-based cache tier sitting in front of them. Cached SSD capacity drives, on the other hand, receive write-only caching, since SSD read latency is already low enough on its own that adding read caching in front of it would provide little additional performance benefit while still consuming cache capacity and adding unnecessary wear to the cache devices.
Combining these two distinct behaviors together, the documented default outcome for this specific three-tier deployment is NVMe-only caching devices with a combined cache behavior of read/write for the HDD capacity tier and write-only for the SSD capacity tier.

When a Storage Spaces Direct cluster is deleted and the underlying physical disks still carry a leftover, now- orphaned storage pool from the previous deployment, Windows Admin Center and Server Manager reports a storage error because the redeployment process expects to find raw, unclaimed disks rather than disks that already belong to a read-only, non-primordial pool with no valid virtual disks in it. Removing Pool1 clears out that stale pool object and its metadata so the disks are no longer associated with a defunct storage pool.
Resetting the physical disks in Pool1 goes a step further by wiping the disk-level Storage Spaces Direct metadata (partition tables, pool membership records, and cache/journal reservations) directly from each physical disk, which is necessary because simply removing the pool object does not always clear every low- level artifact that the new S2D deployment ' s disk-eligibility checks scan for. Together, removing the pool and resetting the physical disks fully clear the stale configuration so the disks present as clean, eligible candidates for the new deployment. Making Pool1 writable would only allow administrative changes to a pool that needs to be removed entirely, not repaired. Creating a new storage pool from the same disks while the stale pool and its metadata still exist would fail or simply recreate the same conflict. Clearing the cluster configuration from each node addresses cluster membership, not the disk-level storage metadata causing the reported storage error.

DC3 is a domain controller for the existing corp.fabrikam.com AD DS forest, deployed into Vnet1 in Azure.
Domain controllers rely on AD-integrated DNS for locating other domain controllers, SRV records, and replication, so DC3 needs to host that zone itself, which requires installing the DNS Server role on DC3 and letting it host an AD-integrated copy of the corp.fabrikam.com zone. Azure ' s default virtual-network- provided DNS resolver has no awareness of AD DS records and cannot host or serve an AD-integrated zone, so every resource in Vnet1 that needs to locate the domain (including Azure VMs and DC3 ' s own replication partners) must be pointed at real DNS servers instead. That is accomplished by configuring the custom DNS Servers setting on Vnet1 to reference DC3 (and, once available, DC1 and DC2), which makes all Azure VMs in the virtual network resolve names through the AD-integrated DNS infrastructure rather than the platform resolver. An Azure DNS zone (public) or an Azure private DNS zone are separate, unrelated Azure-native DNS services that cannot host an AD-integrated zone and are not used for this scenario. Installing the DNS Server role on DC3 and pointing Vnet1 at it together satisfy the requirement.