Search Our Database

Introduction:

For NovaGPU users who utilize our dedicated GPU as a Service platform for intensive operations like training, fine-tuning, and deploying AI models, efficient resource management is paramount to achieving optimal cost efficiency. High-demand computing resources, whether through public cloud hosting, private cloud hosting, or specialized GPU servers, generate continuous hourly charges.
The need to suspend these powerful resources arises when the NovaGPU instance is temporarily not used—a common occurrence during idle periods between training epochs, scheduled maintenance, or project phases. The ability to shelve these instances is the mechanism to conserve valuable cloud credit.

Prerequisites:

• Access to the Customer Portal.
• The user must have at least one GPU instance.
• Ensure Shelve Action Doesn’t Affect Normal Operations

Step-by-step Guide:

Step 1: Access NovaCloud Services

• After logging into the Customer Portal, locate the left-hand navigation menu. Click NovaCloud.

Step 2: Navigate to GPU Instances

• From the expanded NovaCloud menu, locate and Click GPU Instances.

Step 3: Open the Actions Menu

• Identify the specific active GPU instance you intend to stop.
• On the row corresponding to that instance, Click Action. A dropdown menu containing management options will appear.

Step 4: Select the Shelve Button

• From the available actions displayed in the dropdown menu, click “Shelve” option.

Step 5: Confirm the Shelving Action

• Upon selecting “Shelve,” a confirmation prompt will appear on your screen.
• Confirm the shelve action as requested in the dialog box.

Step 6: Acknowledge the Queue Notification

• Once the shelving action is confirmed, the instance is placed in a queue for processing.
• Click “Okay” to close this notification and allow the process to finalize.

Step 7: Verification of Shelved Status

• The process is now complete. You can monitor the instance list to verify that the GPU Instance status has successfully updated to “Shelved”.

Conclusion:

For NovaCloud users who rely on IPSERVERONE to manage their essential, high-performance cloud infrastructure, particularly NovaGPU – GPU as a Service, the Customer Portal provides a vital mechanism for controlling the operational states of their specialized GPU Instances.
This functionality, which is accessed directly through the NovaCloud dashboard, is specifically designed for situations where a GPU Instance is temporarily not needed for computationally demanding operations, such as training, fine-tuning, and deploying AI models.
Shelving an instance is a strategic action that effectively pauses its activity, while crucially preserving its current state and configuration, which is highly beneficial for resource management and optimizing cloud expenditures. This comprehensive approach ensures that users can effectively manage their high-cost GPU resources, adapting to fluctuating demands without the intensive process of completely de-provisioning and later recreating instances, thus maintaining operational efficiency and flexibility.