For any non-Azure machine, whether on-premises or hosted in another cloud, to be used as a source for an Azure Monitor Agent-based Microsoft Sentinel data connector such as the Windows Firewall connector, the machine must first be onboarded to Azure Arc as a connected machine. Installing the Azure Connected Machine agent on Server1 is precisely the action that performs this onboarding: it registers Server1 as an Azure Arc-enabled server, giving it an Azure resource identity that can then have extensions such as the Azure Monitor Agent deployed to it and associated with a data collection rule. This Azure Arc onboarding is the documented prerequisite step described in Microsoft ' s guidance for connecting Windows-based, non- Azure servers to Sentinel ' s AMA-based data connectors, and once Server1 is Arc-enabled, the Azure Monitor Agent extension and a data collection rule targeting the Windows Firewall log source can be deployed to it to complete the log collection pipeline. Because installing the Azure Connected Machine agent is exactly the required first step for enabling this scenario, this solution meets the goal.

Explanation:
Older logs overwritten: The Performance Counter properties. Stop below 500 MB free: The Data Manager properties.
The individual Performance Counter data collector inside a Data Collector Set has its own Properties dialog with a File tab that includes an Overwrite option, and enabling that option causes each new logging run to overwrite the previous log file in place rather than creating a new, separately numbered file each time -- this is exactly the control needed to make older performance counter logs get replaced by newer ones. Disk-space- based retention, on the other hand, is governed at the level of the Data Collector Set as a whole through its Data Manager properties, which expose a Minimum Free Disk Space setting, alongside Maximum Folders and Maximum Root Path Size settings, that Windows evaluates against the log directory ' s volume to manage or halt further data collection once free space drops below the configured threshold. Because these two behaviors live on two entirely different property pages -- one scoped to the individual counter ' s file settings and the other scoped to the Data Collector Set ' s overall data management -- the Configuration properties option, which governs sample interval and duration settings, does not control either of these two requirements.

Explanation:
1. Create an Azure Recovery Services vault. 2. Create a Hyper-V site. 3. Install and register Azure Site Recovery Providers. 4. Create and associate replication policies.
Setting up Azure Site Recovery for a Hyper-V cluster that is not managed by Virtual Machine Manager follows a fixed infrastructure-preparation order. An Azure Recovery Services vault must exist first, since it is the container that every later step registers against. A Hyper-V site is then created in that vault to represent the on-premises Hyper-V environment, because a Hyper-V host or cluster must belong to a site before its hosts can be registered against Azure Site Recovery. The Azure Site Recovery Provider is then installed on the Hyper-V hosts (Cluster2 ' s nodes) and registered to that site/vault, which is what actually connects the cluster nodes to Azure Site Recovery. Only once the hosts are registered can a replication policy be created and associated with the Hyper-V site, since a policy has nothing to attach to until the site and its registered hosts already exist. Installing Azure Connected Machine agents is unrelated to Hyper-V VM replication -- that agent is used for onboarding servers to Azure Arc -- and is not part of this workflow. Therefore, the correct order is: create the vault, create the Hyper-V site, install and register the Providers, and create and associate the replication policy.

Hyper-V Replica ' s Additional Recovery Points setting, which is configured on the Configure Recovery History page of the Enable Replication wizard when replication for VM1 is first set up, lets an administrator specify that a series of hourly recovery points, for up to a maximum of 24 hours of history, be retained on the replica server alongside the single latest, most up-to-date replica of the virtual machine. Because the requirement in this scenario is to be able to restore VM1 to a specific state from as far back as eight hours ago, the Enable Replication wizard ' s recovery points configuration must be used and set to retain at least eight hours of Additional Recovery Points, since without this setting enabled with a sufficiently long retention window, only the latest replica state would be available and no historical point-in-time restore within that eight-hour window would be possible. Enabling automatic checkpoints on VM1 ' s own properties is unrelated to Hyper-V Replica ' s separate point-in-time replica history mechanism and would not create restorable recovery points on the replica server at all, NUMA spanning is a memory-configuration setting on the Hyper-V host with no bearing whatsoever on replication or recovery points, and modifying the replication frequency only changes how often changed data is sent from Host1 to Host2, not how many historical recovery points are retained once it arrives, so configuring recovery points from the Enable Replication wizard is the only option that actually addresses this requirement.

Data Deduplication relies on a set of scheduled jobs to keep a volume ' s chunk store efficient. The optimization job is the one that scans eligible files and replaces duplicate data with references into the chunk store, and it is what produces the initial space savings, but it does not reclaim space that becomes unused because files were later modified or deleted; over time, chunks that are no longer referenced by any current file simply accumulate in the chunk store, still consuming disk space, unless something specifically reclaims them. The garbage collection job is the dedup job dedicated to that reclamation: it identifies chunks in the chunk store that no longer have any live references (because the files that used them were deleted or overwritten with different content) and removes them, directly reducing the space consumed by data that used to belong to recently modified or deleted files. Garbage collection is configured, like the other deduplication jobs, through Set-DedupSchedule, which lets you define or adjust the job ' s schedule and priority. The scrubbing job instead checks the integrity of the chunk store and repairs corruption using redundant copies, and InputOutputScale/IO settings tune deduplication job performance rather than controlling what gets reclaimed; neither addresses the stated goal. Configuring a garbage collection job with Set-DedupSchedule is therefore the correct action.