Understanding Azure Costs: A Practical Guide[Redis] Part 1

This article is the first in a series where I share my experiences in calculating Azure costs for various services. Through real-world use cases, we will uncover hidden costs and determine accurate pricing for Azure resources. This series will help you understand and predict cloud costs more effectively for your specific workloads.

Use Case. Estimating the Cost of Azure Redis Cache

Scenario

I plan to deploy Azure Cache for Redis (Standard Tier) in the East US region with the following specifications:

• Storage: 100,000 items, each 1KB in size (~100MB total).
• Item Expiry: 24 hours (items automatically expire after this duration).
• Read Requests: 40,000 reads per minute (~667 reads per second).
• Write Requests: 100,000 writes per day (~1.16 writes per second).

To estimate the cost of using Azure Redis (Standard Tier) for the described workload in the East US region, let's break down the pricing based on the usage pattern.

1. Storage Requirements

• Data Storage: Plan to store 100,000 items, with each item being 1 KB.

  • Total storage needed:
    100,000×1 KB=100,000 KB=100 MB

• TTL (Time to Live): Each item has a TTL of 24 hours, so the data will automatically expire, maintaining a constant size of 100 MB.

2. Throughput

• Read Requests

  • We are making 40,000 read requests per minute, which totals to:
    40,000×60=2,400,000 (2.4M) requests per hour
    2,400,000×24=57,600,000 (57M) requests per day

• Write Requests
  • We have 100,000 write requests per day.

3. Redis Instance Sizing

Based on our usage

• For 100 MB of data, a Standard C0 instance (250 MB of memory - $40.15/month) would be sufficient. However, considering the high read throughput, a larger instance like Standard C1 (1 GB memory) may be more suitable.
• The Standard C1 instance can handle moderate throughput and is priced at approximately $100.740/month in the East US region​.

4. Estimated Cost Breakdown

The main cost components include:

• Redis Cache Instance Cost

  • Standard C1 in East US: ~$100.740/month.

• Network Egress Data Transfer (if our application reads data from Redis to external clients):
  • Azure does not charge for data ingress (data written to Redis).
  • For data egress, the first 5 GB per month is free, and subsequent usage is billed at approximately $0.087 per GB.

Given our workload

• Read Traffic: Assuming the average response size is 1 KB per read:
  • Total egress per day: 57,600,000 requests/day×1 KB/request = 57,600,000 KB = 55 GB/day
  • Monthly egress: 55 GB/day×30 = 1,650 GB/month
  • Estimated egress cost: 1,645 GB−5 GB free = 1,640 GB×0.087 = 142.68 USD

Total Monthly Cost Estimate

• Redis Instance (standard C0): $40.15/month.
• Estimated Total: ~$243.42/month. (standard C1)
• Estimated Total: ~$182.83/month. (standard C0)
• Redis Instance (standard C1): $100.740/month.
• Network Egress Cost: ~$142.68/month (if applicable).

Considerations

• If most of our requests are internal (within the same Azure region), network egress charges may not apply, significantly reducing the cost.
• If our workload scales up, we might need a higher-tier Redis instance, which will increase the cost.

If our backend services and Azure Redis instance are both deployed in the same Azure region (East US), then data egress charges will not apply.

Azure does not charge for data transfer between services that are within the same region. This means that all the read and write requests from your backend to your Azure Redis Cache will be considered internal traffic and will not incur additional costs for data transfer.

Updated Cost Breakdown

Given that you won't have Egress charges, our estimated cost would be:

• Azure Redis Standard C0 instance: approximately $40.15 per month.
• Azure Redis Standard C1 instance: approximately $100.740 per month.

So, our total estimated monthly cost will range from $40.15 to $100.740 based on the Tier. This covers your described workload of handling 100K stored items, 40K read requests per minute, and 100K write requests per day without additional egress costs since everything is within the same Azure region.

Conclusion

Accurately estimating Azure costs requires more than just using the Pricing Calculator. It involves analyzing data transfer, replication, and monitoring fees. In this use case, the standard pricing model covers most costs, but additional expenses may arise from egress and logging.

This is the first article in my series. Stay tuned for more real-world Azure cost estimation use cases, including CosmosDB, Event Hub, Function Apps, Application Insights, and SignalR.