Performance‑Troubleshooting

General

This Wiki article is designed to help you identify and narrow down potential performance issues at the OS level.
At the end of the article, you will also find tips on how to best submit performance-related inquiries to our support team.

Onprem- vs. Cloud-Server

Servers operated on-premises at a customer's site typically exhibit relatively low utilization rates.
Consequently, multiple applications or additional Windows Server roles and features are frequently installed and run in parallel on these systems.
Cloud servers, by contrast, offer highly granular scaling capabilities; as a result, many customers allocate the same—or even fewer—resources to their cloud instances compared to their on-premises systems.
However, it is essential to bear in mind that these are shared infrastructures.
In other words, multiple customers share the same physical resources.
For this reason, the performance of a local server cannot be directly compared to that of a cloud server.
Particularly in the context of terminal servers, we therefore recommend avoiding having too many employees work simultaneously on a single instance.

DATEV-Applications

Experience has shown that one vCore per DATEV user should be allocated on the VM.
Therefore, in case of doubt, we recommend opting for the standard CPU—featuring more vCores but a slightly lower clock speed—to ensure that sufficient CPU resources are available for all DATEV users.

Finding the Cause of Performance Problems

Gathering Information

To identify the root cause of the existing performance issues, all relevant information should first be gathered:

• On which VM within the cloud package does the behavior occur?
• How long has the problem persisted?
• How exactly does the reduced performance manifest itself?
• Have there been any changes at the operating system level? (e.g., Windows updates, software installations, or uninstallations)
• How many employees are working simultaneously on the affected VM?
• Are there any dependencies on other VMs within the same package?
• Which type of VPN is being used?
• Via which protocol do users access the VM?
• Which antivirus software is active on the VM?
• Are DATEV applications being used?
  • If so: How many employees are working with them simultaneously?

Antivirus Protection

We frequently receive reports that, in addition to the integrated Windows Defender, users are deploying a secondary antivirus solution.
In such cases, it is absolutely essential to ensure that both security solutions are correctly configured to prevent them from blocking one another or causing unintended side effects at the operating system level.

If performance issues arise during RDP sessions, you should verify whether exceptions for RDP connections have been configured within the antivirus software currently in use.
The absence of such an exception can noticeably impair performance.

Check VM Resources Using the Resource Monitor

To identify the root cause of the performance issues, the next step is to check the VM's resource utilization directly at the operating system level.
To do this, start by using the Windows Resource Monitor, which you can access—for example—via Windows Search.

The Resource Monitor features dedicated tabs for the most critical system resources:

• CPU
• Memory
• Disk
• Network

These tabs provide detailed information regarding current resource utilization.

CPU

In the CPU tab, you can view all processes that are consuming computing power.
Pay close attention to processes that generate significantly higher CPU load than others, and consult with your administrator to determine whether this behavior is normal.
Possible causes:

• a process is frozen or malfunctioning
• the VM has been allocated too few vCores

Additionally, you can check whether the CPU is operating at its maximum clock speed.
For example, if Windows Power Saver mode is active, the CPU is throttled—which results in performance degradation.
(Note: On Hyper-V hosts, power-saving modes are disabled by default.)

RAM

On the "Memory" tab, you can view the VM's RAM usage.
Pay close attention to processes exhibiting unusually high memory consumption.

In the upper section, you will also find the number of "Hard Faults."
A high number of hard faults associated with a specific process may indicate that:

• the VM has been allocated insufficient RAM
• or that a memory issue exists on the host

If you suspect host-related issues, please contact Support-Ticket_erstellen|TERRA CLOUD Support.

Additionally, you should keep a close eye on the overall RAM utilization.
The average utilization should not exceed 80%.
Experience shows that values ​​above this threshold lead to bottlenecks in memory management and increase the likelihood of hard faults as well as performance issues.

Storage

On the **Disk** tab, you can view—among other things—the **vDisk response times** per process.
It is best to sort this column in **descending** order.

Important notes:

• Response times **> 10–15 ms** indicate a potential issue.
• The **"Time with max. activity"** metric should **not consistently exceed 80%**.

If the vDisk is under sustained overload, this may indicate that **Performance Storage** is required instead of Standard Storage.
Typical accompanying symptoms include disk errors in the Windows Event Log (IDs **153** or **11**).

Network

Translations:Performance‑Troubleshooting/21/en

On the Network tab, you can view current network usage, which can also be sorted by process.
If a single process consumes an unusually large amount of bandwidth, this can negatively impact the performance experienced by other users.

A typical example is TERRA CLOUD Backup, which can consume a significant portion of the available bandwidth at certain times—unless configured otherwise.

Create Support Ticket

If the cause of the performance issues cannot be identified at the operating system level, please contact Support.
To do so, use either the TERRA CLOUD Technical Center or send an email to support@terracloud.de.
To process your request, we require the following information:

• the name of the affected VM
• the order number
• the answers to the questions listed above