For Hyper-V hosts, the recommended baseline is to let Windows manage the host page file size automatically. That is the documented recommendation after installing the Hyper-V role.
For dense virtualization hosts, page file sizing is not based on a single fixed formula. It depends on installed RAM and workload commit demand. A common starting point in Windows is about 1.5 × installed RAM when no better workload-specific data is available, but on server systems the better goal is to have enough physical RAM so the page file is not heavily relied on.
For Hyper-V-specific stability:
- Leave the host page file system-managed unless there is a measured reason to change it.
- Hyper-V documentation recommends allowing the system to manage page file size.
- This is the safest default for shutdown and memory-pressure scenarios.
- Measure actual host memory headroom before changing anything.
- In Performance Monitor, check Hyper-V Dynamic Memory Balancer – Available Memory.
- If this value is consistently low, the host is under memory pressure and page file changes alone will not fully solve stability issues.
- Check whether the VMs themselves are under memory pressure.
Inside Windows guests, use these counters:
- Memory – Standby Cache Reserve Bytes
- Memory – Free & Zero Page List Bytes
- Memory – Pages Input/Sec Suggested thresholds from Microsoft:
- Standby Cache Reserve Bytes + Free & Zero Page List Bytes should be 200 MB or more on systems with 1 GB RAM.
- The same combined value should be 300 MB or more on systems with 2 GB or more RAM.
- Pages Input/Sec average over 1 hour should be less than 10. If these are poor, increase VM memory sizing rather than relying on paging.
- Use Dynamic Memory where appropriate.
Hyper-V can adjust VM memory dynamically using startup, minimum, and maximum values.
- If a VM has sudden large memory allocations, increasing the guest VM page file size can provide temporary backing while Dynamic Memory responds.
- For restart scenarios, Hyper-V can use Smart Paging only temporarily and only under specific conditions, mainly when a VM is being restarted and there is no available physical memory.
- Smart Paging is not the normal mechanism for general host memory oversubscription.
- Monitor paging behavior instead of sizing blindly.
Useful counters include:
- \Memory\Page/sec
- \Memory\Page Reads/sec
- \Memory\Page Inputs/sec
- \Memory\Page Writes/sec
- \Memory\Page Output/sec
- \Paging File(*)% Usage Sustained high hard page fault rates indicate disk activity that can contribute to poor performance. Correlate these counters with disk performance on the volume hosting the page file.
Practical configuration steps for the host page file:
- Open System Properties.
- Select Advanced.
- Under Performance, select Settings.
- Open the Advanced tab.
- Under virtual memory, configure the page file.
- Prefer automatic/system-managed sizing for the Hyper-V host.
If a manual size must be used:
- Start from workload measurement, not guesswork.
- Use the Windows guidance of 1.5 × RAM as an initial baseline only when no better data exists.
- Validate with Performance Monitor after the change.
- Be aware that undersizing the page file can reduce stability under memory pressure, while oversizing mainly consumes disk space.
For the shutdown problem described, the most practical approach is:
- Ensure the host has sufficient free disk space on the volume used for paging.
- Keep the host page file system-managed unless monitoring proves a different size is required.
- Verify host available memory with Hyper-V Dynamic Memory Balancer – Available Memory.
- Reduce VM memory pressure by correcting VM sizing or Dynamic Memory settings if guests are paging heavily.
This combination is the supported path to improve shutdown reliability and reduce the chance of failures caused by memory and disk pressure.