Scale Up Nerdio Manager for Large Deployments

Scale Up Nerdio Manager for Large Deployments

As your AVD environment scales to hundreds or thousands of session host VMs, along with 10,000+ users, you need to increase the size of Nerdio Manager's default Azure PaaS services to accommodate the large AVD deployment.

By default, Nerdio Manager is deployed with sufficiently large PaaS resources to accommodate a few hundred hosts and a couple of thousand users. See Nerdio Manager Default Deployment Resources and Costs for details.

Azure SQL Database

The Azure SQL database is likely to be the first resource that needs to be increased in size. Change the pricing tier to one with at least 100 DTUs (for example, S3+ or P1+).

Notes:

  • After increasing the pricing tier, monitor the Compute utilization graph on the Overview page. If utilization is consistently between 80% and 100%, gradually increase the number of available DTUs.

  • Changing the Azure SQL database pricing tier is non-disruptive and can be done during production hours.

App Service Plan

If App Service CPU or Memory utilization is consistently very high, increase the size of the App Service Plan (Scale up) to from Nerdio Manager's default of B3 to a larger instance size - up to P3V2. CPU & Memory consumption can be viewed from the app service plan (select the 'nmw-app-plan-*' link displayed on the app service overview tab, on the App Service Plan field).

CPU or memory consistently high indicate the app service must be increased for best performance and functionality:

To change the app service plan, from Nerdio Manager's app service, select Scale up (app service plan) and choose a larger Standard or Production size (Nerdio Manager's default is B3. We recommend starting with S3 or P2v2, but can scale up to P3v2).

Note: Changing the Azure App Service plan size is non-disruptive and can be done during production hours.

Nerdio Manager User Interface API Optimizations

Refer to Advanced App Service Configurations for more information.

In large AVD environments with many host pools or hosts, the time to reload or refresh workspace and pool details, as well as reducing expensive API calls, can be improved by modifying VM request behavior. The following settings are recommended in large AVD deployments:

  • HostPoolVmsConfig:PowerStateRequestBehaviour (value: ForAllVms)

  • HostPoolVmsConfig:RequestBehaviour (value: ByPool)

  • AutoScale:BatchMode (value:TRUE)

  • UI:RecentTasksHours (value: 24)

Note: Changing the App Service configuration settings requires a restart of the App Service and makes the Nerdio Manager portal unavailable for up to 5 minutes while the App Service is restarting. App Service restart has no impact on user's AVD connectivity.

Azure API Limits and Considerations for Nerdio Manager

Large environments may also reach request thresholds that exceed Azure's API limits, which can impact Nerdio Managers ability to successfully maintain the AVD environment. There are several more advanced configuration changes and features within Nerdio, that may be adjusted or turned off if unused, to assist in reducing the number of API requests needed and operate within Azure API limits.

Recommendations for reducing Azure API calls and Nerdio Manager best practices for AVD environment sizing are detailed in Azure API Limits and Throttling Overview.

For additional advanced configuration settings, please refer to Advanced App Service Configurations for more information.

Depending on the size of the environment, you may still reach request thresholds that exceed Azure's API limits. The Azure API Limit Booster feature allows you to grow the total number of API calls by a factor of n, where n is the number of client apps linked for Nerdio Manager to use. We round-robin the applications as Nerdio Manager performs normal operations, distributing the API calls among each application that is linked. In theory, you can use this method to scale the size of your environment indefinitely. See Azure API Limit Booster for details.

Was this article helpful?

0 out of 0 found this helpful
Have more questions? Submit a request

Comments (0 comments)

Article is closed for comments.