Infracost

Happy Wednesday!! Who is coming to FinOps X Barcelona on Nov 11th? Me, Ali and a new Infracost person (!!) will be there - here is what we look like. Come say hi, give me a high 5. We are friendly :) AND I can show you what 3 years of Shifting FinOps Left looks like. Much love, Hassan #finops #finopsxe

Checkout our new e-book about Shifting FinOps Left and what we've learnt in the last 3yrs of building that vision with Infracost! https://lnkd.in/g_GmXZu5

A few months back, I sparked a discussion on why FinOps professionals often avoid strict policies compared to their security counterparts. Since then, I've seen a shift—more engineering leads at Infracost customers are moving FinOps policies into enforcement mode. Just this week, I noticed a customer upgraded all their gp2 EBS volumes to gp3 and enforced this as a policy to prevent any more gp2 volumes from being added. They did the same for io1/io2 volumes. On the EC2 front, they’ve enforced policies to phase out previous-generation instance types, ensuring their to-do list does not grow with new instances. The key to enforcing these policies without frustrating engineers is focusing on new resources. We don’t block changes to already-deployed instances, but when adding new instances with outdated tech, we ask the engineer to upgrade before merging their pull request. Catching these issues early is crucial—prevention is 10x easier than fixing deployed resources and avoiding downtime risks. Looking ahead, I expect more engineering teams will enforce policies for RDS and EKS Extended Support. Upgrading a production database or EKS cluster to newer versions is a massive effort compared to addressing these issues before deployment. Prevention truly pays off. Curious how Infracost can help streamline this process for you? Arrange a demo here: https://lnkd.in/gZ7Bw7qH #ShiftLeft #FinOps #CloudCosts

At the FinOpsX conference there was a Chalk Talk where various practitioners shared their tips beyond basic cost optimizations such as not using gp2... AWS VCP came-up a few times and speaking with Tom Cross he mentioned that developers often just copy-paste VPC resources in Terraform not knowing the cost implications, so it's easy to end-up in a state where there are a lot of duplicates, and hence unnecessary costs💰📈 The underlying problem is that engineers are not aware of FinOps best practices and if you tell them after-deployment, it's time consuming and risky to change network settings. Actually most practitioners in that Chalk Talk were making notes as they don't know all of the these advanced tips either. There's a lot of tribal knowledge in the FinOps community. That's why we go broad and talk to lots of practitioners regularly to get cost optimization tips 😃 Every month we codify these best practices from the community into Infracost, and we just added the following two new ones! 👉 1. Consider avoiding multiple NAT gateways for a single subnet: each NAT gateway costs around $420/year so when possible, reuse existing gateways to avoid unnecessary costs. 👉 2. Consider avoiding multiple VPC endpoints for the same service: Each VPC endpoint costs around $80/year so when possible, reuse existing endpoints to avoid unnecessary costs. Infracost sits in the engineering workflow, at the pull request stage, and points-out the above recommendations as soon as it detects duplicate VPC gateway or endpoint has been created in the code. That's what preventive FinOps looks like; and engineers love it as it enables them to ship cost-efficient code from the beginning, no need to go-back and change things ✅ What's the next FinOps best practice we should codify? #ShiftLeft #FinOps #CloudCosts

Happy Tuesday (13!!) - got a really cool product today! A little while ago, Owen joined us at Infracost. Hit the ground running of course, and today’s product is his creation. It surprises non-developers that engineers can write code, and release it, spend hundreds of thousands of dollars without ever seeing the cost upfront - there is no checkout screen!! 🤯 Well, we used to have a VSCode Plugin, and now we have a JetBrains plugin! See the cost impact of your code as you develop it! It is now live, check it out, share the love with your friends: https://lnkd.in/g3WK5EAf

The best engineering teams we see at large enterprises use modules to package their infra-as-code (e.g. Terraform) into components that can be re-used and shared across their organization. Often there will be Platform or DevOps teams that create and maintain the modules; the modules are then used by various development teams💡 This setup creates an interesting problem when it comes to FinOps tags: sometimes tags need to be fixed in the module itself by adding a variable that can be set by development teams. However, development teams aren't usually aware of this so they just use the module and move on to their next sprint task 🏃♀️ Meanwhile, even if Platforms/DevOps team notice this problem, they cannot fix it themselves as they can just add the tag variable, but the development teams need to set the tag value itself. Though by now, the development team has moved on to another sprint task... That frustrates the FinOps team as there are untagged or mistagged resources; and they need to chase-up two different teams and get them to work together to fix the problem! No wonder why "encouraging engineers to take action" is the number one problem faced by FinOps teams 😓 We've been discussing this problem with our enterprise customers for a while and we just released a feature in Infracost to solve the root cause of this problem! Here's how it works: 1. The FinOps team take 2mins to define their tagging policy in Infracost, including a custom message that they can use to link to their Confluence or wiki page where engineers can learn more about why tagging is very important. 2. Infracost will fetch and process modules when it's parsing the development team repos. Whenever a code change is made (a pull request is opened), Infracost will let the engineer know exactly which resource code blocks have tagging issue, and importantly which modules are provisioning them. It will literally link them to the file and line numbers that have the issue so they can be fixed quickly. 3. Infracost will also validate the tag values and provide suggestions when there are typos or closely matching values (we see that a lot). This is the fastest possible feedback loop and it's all preventative care; it happens before code deployment! That is where it's easiest to fix tagging issues. That's what Shift FinOps Left looks like: the right people, getting the right information, at the right time, in the right place 🚀

I'm doing a webinar about shifting FinOps left with Stephen Old and Rich Gibbons on the 30th of July - let me know what topics/questions you'd like me to cover! #CloudCosts #FinOps #ShiftLeft

It's been 14 years since AWS started supporting resource tags and FinOps teams are still complaining about "our tags are a mess" - why? Everyone knows that without good tags, it's very difficult to create a culture of cost ownership, showback or optimizations. So why are we still wasting so much effort in FinOps and Engineering teams on tagging issues? I think it's because we've not solved the root cause. We treat tagging as a post-deployment task, which makes it a never-ending frustration for FinOps teams as they constantly have to chase Engineering teams to "fix tags". That's also frustrating for Engineering teams as they have to go back and change something they just released. The process is broken, it's not efficient. Shift Left is how we have solved the root cause and the results are mind-blowing 🤯 See the attached screenshots from an Infracost user with 100+ code repos. They went from 9% resources tagged correctly to 90% in 3months. They took 2mins to define their tagging policy in Infracost's dashboard, and engineers starting seeing the exact file+line number they needed to fix as soon as they made a code change. That's the right info (tag typo) in the right place (GitHub) at the right time (as soon as the pull request is opened, BEFORE it is merged and deployed). That's what Shift FinOps Left looks like. It removes the frustrations and gets results quickly.

At the FinOpsX keynote, J.R. Storment mentioned that "FinOps has shifted left" ⬅️ We've been building and enabling that since 2021 with Infracost so it's awesome to see the community recognize the shift in the industry.

Happy Monday 🎉 FinOps X San Diego, we are coming for you! June 19th to 22nd. Here are what our faces look like, come up to the booth and give us a high 5 please. We are nice, and we don't like being alone 🤗