How to manage AWS cost Optimization

by

Cloud bill is an issue for some companies. During periods of time where survive through growth is the only priority, some secondary issues are created.

When growth is not the only priority and profits is the priority there is not question that expenditures have to be reviewed. For Tech business, the cloud bill is part of the equation.

Cloud solutions offer cost management services

The major cloud players offer solutions to track, analyze and improve the cloud bill. There are many details and aspects that need to be taken into account.

The point is how to pay attention internally to this priority, the basis.

Basic checklist and questions for reviewing the cloud costs

The revision of costs should be systematic, You should have a process to analyze the data in a consistent way, adding on top aleatory researches to find new potential gaps.

Below, some steps, questions or points that could help you:

1.- Define a strategy:

  • Define the purpose you have,
  • How you want to tackle this issue,
  • Assign the right people in the engineering and finance units.
  • Work on the ground rules and style you want to approach this. Prepare for rejection and other hurdles.
  • Prepare how this new “priority” fits into the rest of “top priorities” you are pushing.

2.- Launching the program

  • Define the proper business KPIs that can bridge the conversation between Engineering and Finance.
  • Review how the creation of assets are done:
    • Who can do it,
    • How it’s done
    • How resources are tagged
    • etc.
  • Define a basic process and improve it as you learn.
  • Reduce cost is an iterative process that implies understanding the end to end decisions around design and this is something it requires different loops till you find out answers.

3.- Improve your data quality and process:

  • Tag the assets by value stream team.
  • Tag if it’s a permanent asset or a temporary asset.

4.- Improve the way you analyze the infrastructure:

  • Have pareto principle in mind: 80% of the issues are in 20% of causes.
  • Review the data transfer routes.
  • Are you provisioning in advance the services you are going to use?
  • Reduce ephemeral instances to the minimum.
  • Teams consuming large amount of resources have to monitor them during these periods. For instance a Machine Learning training execution that is handling terabytes and high amount of CPU.
  • Automate termination of ephemeral instances or mark them with an expire date.

5.- Review the architecture and the design

  • Educate the engineers about the “cost” issue.
  • Pay attention to multi-cloud solutions and the costs of data transfers.
  • Work with them about how design and architecture can be adapted to be more effective.
  • There are many specific tips and ways to reduce cost from the design point of view. This is just an example on AWS EMR.
  • Who is reviewing the well-architecture Tool from AWS?
  • Who is reviewing the Cloud Financial Management solutions? (similar in MS-Azure).
  • Look for duplications, yes, duplications, be sure there are not redundant assets without a business reason.
  • Think about how you can add predictability to the expenses.

Closing

This list of steps and considerations are just the starting point. The coordination between engineering and finance is critical here to unlock the major issues without eroding the business.

Leave a Comment