I would like to suggest adding **I/O Delay (%)** as a first-class resource usage indicator in the Proxmox VE Web UI.
Currently, Proxmox provides useful real-time information for resources such as CPU usage, memory usage, disk activity, network traffic, and other metrics. However, **I/O delay is particularly valuable when diagnosing storage-related performance problems**, and it would be extremely useful to see it directly in the Web UI.
### Proposed feature
Add an **I/O Delay (%)** column/indicator for every VM and container, showing the percentage of time that the guest is being delayed while waiting for I/O.
Ideally, this would appear everywhere Proxmox currently displays resource consumption information, including:
* VM/Container summary views
* Node resource/guest lists
* Datacenter views where resource usage is displayed
* Any other Web UI table or panel showing CPU, memory, disk, or similar resource indicators
For example:
| Guest | CPU | Memory | Disk I/O | I/O Delay |
| ------ | --: | -----: | -------: | --------: |
| VM 100 | 12% | 45% | 120 MB/s | 0.2% |
| CT 101 | 8% | 32% | 35 MB/s | 7.8% |
| VM 102 | 25% | 71% | 250 MB/s | 18.4% |
The value should update in real time in the same way as the existing CPU and memory indicators.
### Why this would be useful
I/O delay can immediately reveal performance problems that are otherwise difficult to identify.
For example, a VM may show relatively low CPU utilization while applications inside the VM are extremely slow. Seeing an elevated I/O Delay value would immediately indicate that the guest is spending significant time waiting for storage operations.
It would also make it much easier to identify **which particular VM or container is contributing to storage contention on a node**.
This would be especially useful with:
* Busy ZFS pools
* Ceph storage
* RAID arrays
* Slow or overloaded disks
* Shared storage
* NVMe/SATA/SAS storage under heavy load
* Nodes hosting many VMs or containers
### A useful diagnostic improvement
One of the biggest advantages would be being able to spot the problem **without having to open each guest individually or resort to command-line tools**.
A node could, for example, have 30–50 guests running normally, while only one or two guests are experiencing significant I/O delay. Having the value directly in the resource table would make those guests immediately visible.
An optional enhancement could even be to make the column **sortable**, allowing administrators to quickly identify the guests currently experiencing the highest I/O delay.
### Suggested UI behavior
The implementation could follow the same approach already used for CPU and memory indicators:
* Real-time percentage
* Optional historical graph when opening the guest
* Sortable column
* Consistent availability for both VMs and LXC containers
* Optional color/warning threshold when the value becomes unusually high
For example:
**I/O Delay: 0.2%** → normal
**I/O Delay: 5%** → elevated
**I/O Delay: 20%** → significant storage contention
The exact thresholds could of course be determined by the Proxmox developers.
### Why I think this would be a valuable addition
Proxmox already exposes the underlying information, and administrators can obtain I/O delay through existing monitoring mechanisms. Bringing this metric into the Web UI would therefore turn information that is currently mostly useful to experienced administrators into a **simple, immediately actionable performance indicator**.
It would also complement the existing CPU, memory, disk and network metrics rather than duplicating them.
**In short: if Proxmox can tell us how much CPU and memory a guest is using, it would be extremely useful to also tell us how much time that guest is spending waiting for I/O.**
This relatively small UI addition could make troubleshooting storage bottlenecks considerably faster, especially on heavily consolidated Proxmox nodes.
Anyone find this useful?
I opened a request here:
https://bugzilla.proxmox.com/show_bug.cgi?id=8115
Currently, Proxmox provides useful real-time information for resources such as CPU usage, memory usage, disk activity, network traffic, and other metrics. However, **I/O delay is particularly valuable when diagnosing storage-related performance problems**, and it would be extremely useful to see it directly in the Web UI.
### Proposed feature
Add an **I/O Delay (%)** column/indicator for every VM and container, showing the percentage of time that the guest is being delayed while waiting for I/O.
Ideally, this would appear everywhere Proxmox currently displays resource consumption information, including:
* VM/Container summary views
* Node resource/guest lists
* Datacenter views where resource usage is displayed
* Any other Web UI table or panel showing CPU, memory, disk, or similar resource indicators
For example:
| Guest | CPU | Memory | Disk I/O | I/O Delay |
| ------ | --: | -----: | -------: | --------: |
| VM 100 | 12% | 45% | 120 MB/s | 0.2% |
| CT 101 | 8% | 32% | 35 MB/s | 7.8% |
| VM 102 | 25% | 71% | 250 MB/s | 18.4% |
The value should update in real time in the same way as the existing CPU and memory indicators.
### Why this would be useful
I/O delay can immediately reveal performance problems that are otherwise difficult to identify.
For example, a VM may show relatively low CPU utilization while applications inside the VM are extremely slow. Seeing an elevated I/O Delay value would immediately indicate that the guest is spending significant time waiting for storage operations.
It would also make it much easier to identify **which particular VM or container is contributing to storage contention on a node**.
This would be especially useful with:
* Busy ZFS pools
* Ceph storage
* RAID arrays
* Slow or overloaded disks
* Shared storage
* NVMe/SATA/SAS storage under heavy load
* Nodes hosting many VMs or containers
### A useful diagnostic improvement
One of the biggest advantages would be being able to spot the problem **without having to open each guest individually or resort to command-line tools**.
A node could, for example, have 30–50 guests running normally, while only one or two guests are experiencing significant I/O delay. Having the value directly in the resource table would make those guests immediately visible.
An optional enhancement could even be to make the column **sortable**, allowing administrators to quickly identify the guests currently experiencing the highest I/O delay.
### Suggested UI behavior
The implementation could follow the same approach already used for CPU and memory indicators:
* Real-time percentage
* Optional historical graph when opening the guest
* Sortable column
* Consistent availability for both VMs and LXC containers
* Optional color/warning threshold when the value becomes unusually high
For example:
**I/O Delay: 0.2%** → normal
**I/O Delay: 5%** → elevated
**I/O Delay: 20%** → significant storage contention
The exact thresholds could of course be determined by the Proxmox developers.
### Why I think this would be a valuable addition
Proxmox already exposes the underlying information, and administrators can obtain I/O delay through existing monitoring mechanisms. Bringing this metric into the Web UI would therefore turn information that is currently mostly useful to experienced administrators into a **simple, immediately actionable performance indicator**.
It would also complement the existing CPU, memory, disk and network metrics rather than duplicating them.
**In short: if Proxmox can tell us how much CPU and memory a guest is using, it would be extremely useful to also tell us how much time that guest is spending waiting for I/O.**
This relatively small UI addition could make troubleshooting storage bottlenecks considerably faster, especially on heavily consolidated Proxmox nodes.
Anyone find this useful?
I opened a request here:
https://bugzilla.proxmox.com/show_bug.cgi?id=8115