In partnership with
AI is doing great stuff, but costing companies a hell part of a budget. Although they don’t care that much, today we’ll bring you some strategies to bring those AI costs down. After all, you are a FinOps-aware person, you should care, right?
Right?
What also happened this week:
Several AI Cost Reduction Strategies
Pay Less for Serverless
Reduce AWS Route 53 Costs
Mention of Honour: FinOps Event for XE Recap!
Quick Wins with interesting articles. Quiet week from providers’ news.
Professional Spotlight with one of the MVPs in FinOps
Quiet week overall in announcements, yet a lot of good tutorial content. Good stuff for learners like me.
👨🏭 Action Item: Read today’s newsletter, and register for our event 🙂
Collaborate with FinOps Weekly
AI
3 Strategies to Reduce Your AI Costs
Google Cloud experts share great tips to manage AI costs better (not only for GCP but in general).
Choose AI projects that help your business. Don’t do AI for the sake of it. Pick projects that solve real problems and fit with your company’s plans.
Understand all the costs of AI. Things like:
Running the AI models
Training the AI
Storing data
Setting up and running apps
Ongoing support
Knowing these costs helps you make smarter choices about which AI to use.
Use FinOps.
The complete article provides tons of value for those who want to learn how to control AI costs today.
In partnership with Envisor
Thinking about building your own (BYO) FinOps tooling?
Full BYO is often more challenging than it seems. While FOCUS eases multi-cloud visibility, what will you do about data enrichment? You’ll need data points like pricing, observability and more. In fact, from AWS alone you will need to pull data from 35 API endpoints in order to build a complete solution.
Research reveals that even after two years, most BYO’ers still rely on third-party CCM solutions. So, how real are those cost savings?
Envisor combines the power and customization of BYO with the convenience of SaaS. It delivers the same powerful features as other leading platforms but with an open data lake architecture and a Power BI front end that provides complete control over your FinOps visualizations.
FinOps Foundation
Guide to AI Cost Tracking
As companies use more AI tools, it’s important to track how much they cost. The guide by FinOps Foundation provides how to do that by focusing on tokens, which are the main way AI services charge for usage. The article suggests three ways to track costs:
Count API requests (simplest but least accurate)
Estimate token usage (more accurate but needs more work)
Use actual token counts (most accurate)
It’s best to store this information in a database. The article provides an example of how to set up the database tables.
Then they explain two ways companies can set up AI systems. Centralized, where all AI goes through one central hub and decentralized, with different teams using AI independently. Centralized is easier for tracking costs, but decentralized can be more flexible.
Lastly, they discuss how to handle costs for companies that buy AI processing power in bulk (called Provisioned Throughput Units or PTUs). This can save money but makes cost tracking more complex. It was a complete guide overall!
Serverless
How to Pay Less for Serverless
Orel Bello works at a company called Melio that uses a lot of serverless computing. He shares several tips to help reduce costs:
Use AWS Lambda Power Tuning to find the right amount of memory for your functions.
Set the right timeout for your functions to avoid paying for unused time.
Put long-lasting resources outside the main part of your function to make it run faster.
Use ARM-based processors instead of x86 to save up to 20% on costs.
Be careful when using Provisioned Concurrency, which keeps functions ready to run quickly.
Don’t use ‘sleep’ in your functions. Instead, use Step Functions to manage waiting periods.
Consider using the AWS Compute Savings Plan to get discounts on long-term usage.
Be smart about logging. Only log what’s necessary and consider using cheaper storage options.
Use VPC Endpoints to connect to AWS services more efficiently.
More inside, enjoy the complete article!
AWS
7 Ways to Reduce Your AWS Route 53 Cost
Amazon Route 53 is a service from AWS that helps manage domain names and direct internet traffic. While it’s useful, it can get expensive. Here are some ways to lower the costs:
Combine hosted zones: Instead of having separate zones for each subdomain, put them together in one zone.
Use alias records: These don’t cost money for lookups if they point to AWS services.
Reduce DNS queries: Set longer cache times so DNS information stays saved longer.
Be careful with advanced routing: Only use special routing for important applications.
Take advantage of free offers: New AWS users get some free Route 53 usage for the first year.
Limit health checks: Only use these for the most important services.
Think about other DNS providers: Sometimes, other companies might be cheaper for what you need.
By using these tips, you can keep Route 53 costs down while still having a good system for managing your internet traffic.
Together with FinOpsly
AI Copilot for FinOps
End-to-End FinOps.
Data at the speed of thought.
FinOps Empowered by AI.
FinOpsly’s AI-powered platform helps optimize your cloud cost and performance by understanding the unique complexities of your cloud infrastructure.
Multicloud visibility, optimization, and continuous improvement are all in the most intuitive enterprise product that you’ve ever used.
Adam from FinOpsly
🎖️ Mention of Honour
Event: FinOps XE Recap 2024
We are hosting a FinOps XE Recap on the 21st of November! We’ll be gathering top talent from the FinOps Spanish-speaking community in the FinOps XE in Barcelona to talk about their experience in the event. Our elite panelists will be:
Narciso Cerezo, Global FinOps Lead at BBVA
Borja Martínez, FinOps Professional at NTT Data
Anderson Oliveira, Senior Cloud FinOps at AXA
Alfonso San Miguel Sánchez, Lead Cloud Architect at B2 Holdings
Hosted by me and Damian Munafo, we’ll bring you a complete summary of the event. We’ll also have a previous special session, pending to define. Registration is open. Looking forward to seeing you there!
P.S. The Panel will be in Spanish. Open for Sponsorship.
🏆 Quick Wins
Announcements
AWS Billing Conductor (ABC) has added a new feature for its customers. This feature allows users to see information about their Reservation and Savings Plan coverage and use.
AWS Cloud Intelligence Dashboards Framework Update
Oracle and IBM clinch 10-year cross-departmental cloud deal with UK government
Articles & Podcasts
Women in FinOps – Dr. Maneesha Asundi: Sharing Knowledge, Building Empathy – YouTube
Shift FinOps focus to requirements for big impact. Jeremy suggests adding a section about costs to the documents used to plan new software projects.
Borja Martinez shared his experience on the FinOps X. From Madrid to San Diego!
11 Reasons to Skip the Cloud: AWS and other big cloud platforms are not always the best choice for every situation. There are times when it makes sense to use other options.
Solving cloud cost challenges with unit economics at r/FinOps. Measuring cloud unit economics can be tricky.
Professional Spotlight
Dvir Mizrahi
One of FinOps MVPs
Dvir did an excellent job on Wix on the FinOps side. Also contributing to the community and now within Wiv.ai, he continues to be one of the most valuable professionals within FinOps.
That’s all for this week. See you next Sunday!
Let’s Collaborate!
Sponsor this newsletter: Promote your company in this newsletter and reach the Cloud audience that wants to maximize business value and reduce costs.
Collaborate with Us: Our brand expands to more than just newsletters. Podcasts, Posts, Webinars, Events, and any collaboration related to FinOps are available.
FinOps Services: Our partner brand Cloud Cost Control provides FinOps Services you can leverage today. Feel free to request a discovery call.
