Dear PVE Development Team,
As a PVE user, I must express my strong dissatisfaction with the current design and user experience. PVE has been in development for 20 years as an enterprise-grade virtualization platform, and it should have a more mature architecture and user experience. However, the reality is disappointing. Below are several critical issues I encountered while using PVE, which I sincerely hope the team will take seriously and improve fundamentally—rather than leaving users to manually fix problems that should have been resolved.
Suggestions:
Provide a fully functional UI for cluster management, including creation, modification, and deletion, without requiring command-line interventions.
Fix cluster IP recognition mechanisms, ensuring that the system reads the current configuration instead of relying on outdated /etc/hosts data.
Improve the cluster removal process, ensuring a clean uninstallation that does not leave residual data affecting future operations.
Suggestions:
Disable Swap by default (or provide an optimization option) since ZFS inherently discourages Swap usage.
Adjust ARC cache default usage ratio, providing UI options for users to modify it instead of enforcing a strict 50% default allocation.
Improve memory management to prevent premature Swap activation, which negatively affects performance.
Suggestions:
Enhance HA mechanisms to improve stability, preventing unnecessary service disruptions or failed migrations.
Implement a more efficient storage migration mechanism, enabling live migration without performance degradation.
Expand storage backend compatibility, ensuring better integration with enterprise storage solutions.
PVE, as an enterprise-grade virtualization management platform, has been in development for 20 years and should already be a mature and stable product. However, in reality, it is still plagued with design flaws, UI-backend inconsistencies, and inefficient memory management, making it feel more like a "barely functional" open-source alternative rather than a truly competitive enterprise solution.
If PVE genuinely aims to be the open-source alternative to ESXi, then it needs to seriously refine these core functionalities instead of forcing users to continuously struggle with workarounds and patches.
I hope the PVE team will acknowledge these problems and make meaningful improvements in future releases, rather than leaving users disappointed time and time again.
As a PVE user, I must express my strong dissatisfaction with the current design and user experience. PVE has been in development for 20 years as an enterprise-grade virtualization platform, and it should have a more mature architecture and user experience. However, the reality is disappointing. Below are several critical issues I encountered while using PVE, which I sincerely hope the team will take seriously and improve fundamentally—rather than leaving users to manually fix problems that should have been resolved.
1. Severe UI/Backend Disconnection in PVE Cluster Management
- The UI allows easy creation of a cluster, but does not provide a clean uninstallation option, forcing users to resort to command-line removal, which often leaves residual data that disrupts subsequent operations.
- After changing the IP address, PVE still recognizes the old IP from /etc/hosts, instead of the newly configured IP, making cluster reconfiguration extremely difficult.
- The UI and backend logic are completely out of sync, making it feel like an incomplete product.
Suggestions:
Provide a fully functional UI for cluster management, including creation, modification, and deletion, without requiring command-line interventions. Fix cluster IP recognition mechanisms, ensuring that the system reads the current configuration instead of relying on outdated /etc/hosts data. Improve the cluster removal process, ensuring a clean uninstallation that does not leave residual data affecting future operations.2. Poor Swap and Memory Management in PVE with ZFS
- PVE uses Swap even when there is sufficient physical memory, leading to unnecessary performance degradation.
- ZFS ARC cache is overly aggressive, defaulting to 32GB on a 64GB system, significantly impacting other services.
- Despite ZFS being incompatible with Swap, PVE still does not disable it by default, forcing users to manually adjust vm.swappiness or even run swapoff to mitigate the issue.
Suggestions:
Disable Swap by default (or provide an optimization option) since ZFS inherently discourages Swap usage. Adjust ARC cache default usage ratio, providing UI options for users to modify it instead of enforcing a strict 50% default allocation. Improve memory management to prevent premature Swap activation, which negatively affects performance.3. Lack of Enterprise-Grade Features and Poor User Experience
- Cluster HA (High Availability) remains unstable, requiring frequent manual interventions to fix issues.
- No true storage vMotion, making VM storage migration inefficient and restrictive.
- Insufficient backend storage support, making PVE less competitive compared to mature enterprise solutions like ESXi.
Suggestions:
Enhance HA mechanisms to improve stability, preventing unnecessary service disruptions or failed migrations. Implement a more efficient storage migration mechanism, enabling live migration without performance degradation. Expand storage backend compatibility, ensuring better integration with enterprise storage solutions.Final Thoughts
PVE, as an enterprise-grade virtualization management platform, has been in development for 20 years and should already be a mature and stable product. However, in reality, it is still plagued with design flaws, UI-backend inconsistencies, and inefficient memory management, making it feel more like a "barely functional" open-source alternative rather than a truly competitive enterprise solution.
If PVE genuinely aims to be the open-source alternative to ESXi, then it needs to seriously refine these core functionalities instead of forcing users to continuously struggle with workarounds and patches.
I hope the PVE team will acknowledge these problems and make meaningful improvements in future releases, rather than leaving users disappointed time and time again.
