Just the Links - Volume 8

Performance and Price Comparison (on EC2)

A great independent review of the different EC2 types using a Wordpress installation. If anything, it proves out the Graviton processors are a real solution that should be considered given the performance and cost.

Guide to k8s autoscaling for cloud cost optimization

There are two types of autoscaling in k8s: horizontal (add/subtract more nodes) and vertical (resize nodes). The Cluster Autoscaler component can be used to help manage costs, including with a mixed instance strategy that allows you to choose from various instance types.

AWS Lambda@Edge changes billing duration from 50ms to 1ms

Always a good thing.

AWS EC2 Auto Scaling introduces Warm Pools to accelerate scale while saving money

Good for applications that have a long bootstrap process. The instances are pre-initialized but in a stopped state, as opposed to the previous way of having to initialize a fresh instance. The instances will just boot up within about 30 seconds.

AirBnB’s Journey Towards Cloud Efficiency

(Thanks to Chris Aniszczyk for the find) AirBnB’s approach to cost efficiency is a six-part approach:

  • Capacity management: savings plans and RIs
  • Waste elimination: removing unused/underused assets
  • Elasticity: k8s and autoscaling
  • Consumption attribution: cost visibility
  • Consumption monitoring: cost usage trends
  • Demand management: budgeting and forecasting

While not completely comprehensive this is a great post on the benefits in approaching your cost management in this way.

Getting Engineers on Board with Cloud Cost Optimization Part 1/3

Three key takeaways here:

  1. Do a deep dive on end-to-end performance
  2. Understand engineers’ challenges without talking about costs
  3. Help engineers fix stability issues and other performance problems

Getting Engineers on Board with Cloud Cost Optimization Part 2/3

The follow up to part 1 (above), three changes you can make to help achieve better cost performance:

  1. Recognize tools alone won’t solve stability and cost
  2. Models should include costs, capacity, and performance
  3. Have ambitious goals when it comes to yearly targets

Whose Code is it Anyway

This talks about the hard task of assigning ownership to cloud assets. This is something that we all continue to struggle with to some degree.

Predict Your Future and Make Your CFO Happy

Yet another “why tagging is important” story.

AWS S3 Multipart Uploads - Avoiding Hidden Costs from Unfinished Uploads

Creating some lifecycle rules to help manage orphaned multiparts is pretty straight forward and this post goes into some good detail on it.

Outdated Calculus of Cloud Cost Containment

The author makes the case of using APM to help not only monitor and make your cloud workload more efficient.

Serverless Patterns Collection

An awesome collection of serverless workload patterns. Pay for what you use!

Azure adds programmatic pricing API

There are some pretty cool use cases to getting pricing pulled in programmatically: think CICD pipelines that can estimate costs, or alerting mechanisms that notify you when certain prices go down.

EC2 Pricing: The Abridged Guide

A good overview of what EC2 is and how it’s priced - this would be a good post to share with finance or executives as a reference.

Estimate Amazon EC2 Spot Instance cost savings with AWS Glue Databrew, AWS Glue, and Amazon QuickSight

A great step-by-step guide for comparing what your spot costs would be next to on-demand.

Discovering and Deleting Incomplete Multipart Uploads to Lower Amazon S3 Costs

Seems to be a popular topic recently (see above), but this one is authored by AWS. Not as much detail though.

AWS vs Azure vs GCP

A feature and cost-wise comparison of the three major cloud providers. Has some decent charts to reference.

This material provided in association with the FinOps for Engineers newsletter. Like what you see? Consider signing up!