Best Minimal Setup to Simulate Load Balancing in Test Environment

I like both options @Andreas Baumgarten and @Shravan Addagatla , but since the Basic Load Balancer is going to be retired soon, I don't think it will be suitable for future implementations.

Thanks for clarifying the basic requirements for load balancers. If there are no plans to add more servers in the future or if auto-scaling is not enabled, a load balancer is not required as it fails to perform its intended function. Is my understanding correct?

For cost reduction, deallocation is a great option, which I hadn't considered earlier. Thanks for suggesting this! Just out of curiosity, besides these two options, do we have any other options without using third-party services?

Hi @InfraSolutions ,

if you only have one "backend" server a load balancer in front isn't required.

The main task for a load balancer is to "balance" a work load between 2 backend servers for redundancy or load balancing requirements.

(If the reply was helpful please don't forget to upvote and/or accept as answer, thank you)

Regards

Andreas Baumgarten

Hi InfraSolutions

Thanks for the response.

If there are no plans to add more servers or enable auto-scaling, a load balancer becomes unnecessary as it won't distribute traffic across multiple backend instances, making it an extra expense. 

Besides deallocation, here are some other ways to optimize costs in Azure without relying on third-party services: 

• If your workloads are predictable, consider Azure Reserved Virtual Machine Instances for up to 72% cost savings compared to pay-as-you-go pricing.
• For non-critical workloads, Spot VMs offer significant discounts but can be evicted when Azure needs capacity.
• Right-sizing your VMs by selecting a smaller VM SKU can significantly cut costs. Avoid over-provisioning resources.
• Set up automated shutdown schedules for VMs that don't need to run 24/7, reducing unnecessary billing.
• Convert unused premium storage to standard, remove unattached disks, and compress data to minimize storage costs.
• If you have existing Windows Server or SQL Server licenses, use Azure Hybrid Benefit to save on licensing fees.

Thank you for the detailed explanation on how to save costs by selecting the right resources in Azure. Prior to seeing this response, I thought of bringing an additional instance for the load balancer and stopping the VMs to reduce costs, so that only the disk resources would be charged. However, assume if our existing setup is already on a 3-year upfront payment. In this case, should the additional instance for the load balancer be created using the pay-as-you-go option? Or does the upfront payment option allow for a refund for non-usable VMs for a specific period? For example, if we use the additional instance only for 6 months out of 3 years, will the cost for the remaining 2.5 years be refunded or reduced on the next renewal?

for other questions you mentioned above,

• For non-critical workloads, Spot VMs offer significant discounts but can be evicted when Azure needs capacity.

As this is testing environment, we probably won't have any critical workloads, unless we want to test critical tasks that don't work in prod and replicate the same task in non-prod.

• Right-sizing your VMs by selecting a smaller VM SKU can significantly cut costs. Avoid over-provisioning resources.

we already have a lower VM SKU compared to prod

• Set up automated shutdown schedules for VMs that don't need to run 24/7, reducing unnecessary billing.

For pay-as-you-go options, this is applicable. For upfront VMs, is this still applicable?

• Convert unused premium storage to standard, remove unattached disks, and compress data to minimize storage costs.

we are using only HDD storage

• If you have existing Windows Server or SQL Server licenses, use Azure Hybrid Benefit to save on licensing fees.

Yes