By: Iswarya Mohan, Lead ─ Backend Development, Member ─ Cloud Implementation Team, Member ─ Research pod for Canvas Workplace

Introduction

The widespread adoption of Work From Home (WFH) in the IT sector after COVID-19 has made it imperative for businesses to furnish resources. These resources should enhance employee productivity and ensure security. The company is responsible for and must bear the cost of providing the working infrastructure to the employees. Hence, reducing its associated costs is in the company’s interest. The working infrastructure may be provided through traditional means, i.e., through laptops, desktops, etc., which have very high capital expenditures (CapEx). Through the influx of cloud computing, companies now can opt for pay-as-you-go costs, eliminating the CapEx costs. Therefore, it is necessary that they focus on minimizing this cost through cloud cost optimization.

Businesses are migrating to a cloud-hosted infrastructure because of the pay-as-you-go pricing model and low CapEx. Cloud users must look for ways to reduce cloud-operating costs, mainly when employees have a Virtual Desktop Infrastructure (VDI) to perform their day-to-day tasks.

VDI desktop and costing

VDI was born to use the cloud as an alternative to desktops. It provides access to remote machines that have been pre-configured to meet various storage, processing, network, and software requirements. It is known to be the virtual equivalent of a desktop computer, which users can access virtually from any device. VDI environments are centralized and delivered on demand.

VDIs help put workstations in centralized servers and make them accessible on zero-client terminals. This technology is the infrastructure for virtual stations and represents the next step in the virtualization philosophy of workstations.

A VDI based on a virtual machine or server costs more when running, requiring cloud cost optimization. When a device is stopped, it enters the ‘deallocated’ state.  In this state, only the disk and related infrastructure, such as the Network Interface Card (NIC,) get charged, not the machine itself.

There are two types of VDIs:

1. Pooled
   A single device or a collection of devices shared by multiple users. The VDI’s resources are shared among users. A single entity should hold VDI controls. This entity should cross-check with VDI users before turning off the VDI desktop. Users are assigned desktops from a pool of available machines. This is typically used when users only require a generic desktop that cannot be customized.

2. Personal
   A unique device is reserved for a single user. Since the VDI running hours are up to the user’s discretion, VDI controls may be managed by the VDI users themselves. A single user will be given direct access to a desktop. When users need customizable persistent desktops, this is frequently used.

Ideating methods for cloud cost optimization

VDI provides numerous advantages, notably simplified desktop management and fewer hardware purchases. As a result, through cloud cost optimization, capital expenditures and operating costs may be reduced. VDI can reduce desktop costs by up to 50% (as calculated on Azure pricing calculator[i]) when implemented correctly.

1. Provision for the VDI user to turn on and turn off resource

The idea is to delegate cost-cutting responsibility to the VDI resource user. By granting the user of the VDI start and stop authority, there is a greater chance that the VDI will only run when needed. The VDI is started by the user when they begin work. It is stopped by the user at the end of the day without relying on the administrator who oversees VDI management.

2. Admin-controlled automated turn-on and turn-off time based on time zone

The administrator may plan the settings to have set operating hours for the VDI for cloud cost optimization. This is determined by the time zone settings of the VDI users. This specifies when the VDI will automatically turn on/off and when a desktop session will log off.

For Indian Standard Time (IST), for instance, the VDI start and stop timing may be set by assuming that a user begins work at 9:00 am and logs off at 6:00 pm. However, the administrator must be notified if a user wishes to advance or extend the timings so that the necessary adjustments can be made further.

3. User-specified automated turn-on and turn-off time

The above-described method for cloud cost optimization is enhanced by giving the user control over the automated start and stop timing of the VDI. This eliminates the need for manual startup and shutdown of VDI daily. The running hours of the VDI can be easily specified based on their work pattern, making it more convenient for the user.

4. Real-time resource usage monitoring

A feature that tracks VDI usage in real time may be planned for cloud cost optimization. This feature would track when the VDI is turned on, when it is logged in, and when it is turned off. If this information is gathered, it will be possible to recommend the start and stop times to the user without requiring the user to specify them. Furthermore, the user can make an informed decision after reviewing the usage pattern.

5. Cloud cost optimization recommendations based on usage monitoring

The above-mentioned method can be further improved by incorporating an intelligent entity that can suggest how cost optimization can be advanced.

There are two ways to achieve cost optimization:

• Advising the user on when to start and stop their VDI session
• Offering suggestions for how to handle idle time

The method to be used depends on whether the user has a predetermined window during which they take breaks. The user would have the choice of going along with the entity’s recommendation or asking for a specific setting to be applied.

Another way to put a check on the cost incurred by cloud resources is by minimizing the resource configuration. Specifically, if we observe that a resource is not being used to its potential, it is recommended that we downsize the configuration. For example, a VDI has been provisioned, and its CPU utilization has been put to observation for a month. It is seen that the upper limit never crosses 50%. This means that the utility for which the resource is being used for, demands only a part of the current configuration. We can downsize it appropriately which will ensure optimum usage as well as cost savings.

Conclusion

In the era of cloud computing gaining popularity owing to the no CapEx and pay-as-you-go model,  the cost incurred through cloud resources can be minimised to a large extent. This can be achieved through a joint effort by the user and the VDI administrator. The user can be given the control to start/stop the VDI or to set the time for automation scripts to do the job. If this is not viable, scripts could run based on the time zone. Leveraging an intelligent entity to recommend cost-saving strategies is also an option. To optimize cloud costs, it’s important to minimize the running time of VDI. However, it’s also crucial to monitor the actual usage and compare it to the potential usage. This will help ensure that resources are configured to the minimum required level.

[i] Azure Pricing Calculator: https://azure.microsoft.com/en-au/pricing/calculator/