Do you actually need a third-party FinOps tool?

I was recently at a meetup talking to an engineer about FinOps tools and he asked me: “Why would I need an external FinOps tool? Aren’t AWS native tools enough?”

My answer to this is always “It depends”.

Built-in AWS cost tools have come a long way in the past years. And if you’ve never seriously looked at your cloud costs before, definitely start here.

And for a small company that can look through and implement AWS cost recommendations in a couple of afternoons, an external platform is overkill.

However there are some situations where I absolutely recommend third party platforms. By using them, you can take control of your cloud costs in a way that you wouldn’t be able to without them.

Here’s my breakdown:

How big is your cloud environment? Past a certain scale, doing cost optimization manually means that you end up on a treadmill of implementing cost cleanup tasks, like a hamster on a wheel, always moving but never really getting anywhere.

Sure, this is true for small companies too. But waste doesn’t pile up as quickly in small cloud environments. In a large one you might need a team of people doing FinOps full time just to keep on top of things.

This is a waste of engineering time.

If your cloud footprint is big enough that you need a dedicated engineer to do cleanup work, then consider using an automation tool.

Yes, you can build your own automations in house, but then you need the engineer to build and maintain them.

The better solution is to get a FinOps tool to take the load off. And your FinOps team can leave repetitive manual tasks behind, and focus on higher-value work.

Some examples of automation tools include: Wiv, OpenOps, and Adaptive6.

How comfortable are you locking your company into a 3 year compute Savings Plan? If you’re like most people the answer is “not very”.

A lot of companies are missing out on the reduced rates that Savings Plans and Reserved Instances offer because they don’t want to risk over-committing.

I totally get this. Things move so fast these days that who knows where the business will be in 1 year, let alone 3.

However, undercommitting means you keep paying on-demand prices and miss out on discounts. I call this the ‘flexibility tax’ - paying more to stay flexible.

So unless you have extremely stable workloads and an unshakable confidence in the future, I recommend using some kind of rate optimization tool.

Hykell, Archera, and Milkstraw are examples of these. They use different mechanisms under the hood but the end result is the same - you get the reduced rates of SPs and RIs without the risk of being locked in.

I’m a big believer in not only cuttings cloud costs, but understanding the value you’re getting from every dollar you spend. In other words, knowing your unit cost.

However if you want to do this, you need to have your costs properly allocated. If you can’t attribute your costs to a team or project, you can’t know what value it’s bringing.

Unfortunately, tagging is one of the biggest headaches in FinOps.

A small company with a simple cloud setup can realistically set a tagging convention and stick to it doing manual tagging.

However in sprawling companies with 20 departments and a multi-technology stack, this gets difficult if not impossible. Manual tagging breaks down in the face of complexity.

Many FinOps platforms offer “tagless” cost allocation, or some other tagging mechanism that doesn’t rely on AWS-native tags. They can also tag untaggable resources (e.g. data transfer) and deal with the problem of allocating shared services.

Examples include Finout, CloudZero, and Attribute.

These are often larger tools that do a lot more than just cost allocation. They can ingest and unify data from multiple cloud providers and have unit economics as a core capability.

So they are probably overkill for smaller setups, but become more useful the larger and more complex your environment is. Especially if you’re serious about knowing the value you’re getting from the cloud.

Pretty much all the AWS-native cost tools have one thing in common and it’s that they help you retroactively. Compute Optimizer tells you that you already provisioned an oversized instance. Anomaly detector tells you that a cost spike already happened.

I’m not saying there are tools out there that will predict a cost anomaly far into the future - there are not.

But using third party platforms helps you move your mindset from managing costs reactively to doing so proactively. They can help you actively manage your commitments, tie cloud costs to business value, or just automate the boring stuff so your engineers can focus on more forward-thinking projects.

Not everyone is at that stage, and that’s fine. If you’ve never looked at Cost Optimization Hub then start there before you begin shopping for something external.

But if you can relate to what I described then it might be time to add an external solution to your toolkit.