FinOps Anti-patterns
In which Joe talks to some Scots, discusses bad FinOps practices, revels in Craplications, and then tries to like Scottish music.
I was recently interviewed for Cloudsoft’s “Cloudy With a Chance of Brain” podcast. My episode is not out yet but go check them out, I enjoyed myself. Cloudsoft is based in Edinburgh and I had great fun talking with Alasdair and Aled who have such great accents. During our conversation, the question of FinOps anti-patterns came up. I love this topic! Yes, getting things right is important but talking about getting things WRONG is flat-out fun. With that in mind, here are some FinOps anti-patterns:
No Tagging. The first step in the FinOps life cycle is to be informed. If you aren’t tagging, how are you going to be easily informed? Remember that leadership, finance, product managers - all sorts of non-techies need to be able to look and understand what's generating costs, so you can’t just look at code and think ‘All good here!’
Shared accounts. Having one giant dev account that you share with all your apps to do their dev work is like going to a dog park and trying to figure out which turd is associated with which dog. It’s damn near impossible, you get weird looks from everyone and you begin to wonder why you are doing this in the first place.
No centralizing commitment discount management. That's the fastest way to overcommit. This should be taken care of by your centralized FinOps team unless, of course . . .
There is no centralized FinOps team. Perhaps you think FinOps is just common sense and your teams have it under control? This new understanding of the financial impact of what resources are provisioned and how they are used will be absorbed as part of the engineers’ and developers’ day jobs naturally? The problems won’t get that bad or out of hand, right? All bad assumptions. Centralized FinOps teams are a legitimate and necessary function of an enterprise cloud platform. They are also the easiest to justify. My teams have always maintained a positive ROI based on unnecessary expenses we have helped avoid each year.
Those are antipatterns from a FinOps team perspective, but there are also patterns of applications that are not yet fit for the world of FinOps. I once worked for a boss who referred to these as “Craplications”. Are you in charge of a Craplication? It happens to the best of us and sometimes it is completely out of any one person's control. Craplication may sound insulting, please know I mean it lovingly - the term is just so appropriate. Onwards - signs of Craplications:
Nobody on the application team knows how the app actually works. Perhaps all the subject matter experts left the company or took a new role years ago and out of fear, there hasn’t been a major release in years. This team isn’t going to be leveraging DevOp methods, pipelines, or any advanced tech like Kubernetes or serverless. If by chance they are leveraging advanced cloud technologies, the cloud platform teams will hate them for all the tickets and phone calls received when this app has an outage.
It takes 10 hours to recover from a server reboot. This sounds ridiculous but you absolutely know an application like this. If it takes a matrix team effort to reboot an app, it is never going to reboot. This means no rightsizing, no shutting off resources during downtimes, no using any of the basic advantages of being in the cloud. The variable cost model of the cloud will be too rich for this team.
The app team migrated to the cloud-based on a company edict from 3 years ago and they finally got around to it. This app team doesn’t understand why they are migrating to the cloud. They are just doing it because 3 years ago the company started a cloud migration program and they are finally getting around to participating. Two big problems off the bat: the app team likely doesn’t understand what their value proposition is to migrate and secondly, the company has likely iterated and advanced their cloud migration strategy by this point from lift and shifts to rearchitectures and replatforms. It will likely be two more years before that app team hears those messages. They are not ready for the speed of the FinOps life cycle.
I can think of two solutions for Craplications. The first option is adopting the reality of purposeful hybrid environments. This allows applications and developers ready for the cloud to take full advantage of it while also maintaining an environment for the fragile apps that is more cost beneficial to them.
The second option is to replace the application with a SaaS solution. You no longer have to maintain the code, likely get rid of decades of tech debt, and move to a consumption-based cost model.
I’ll open up the comments section, which I normally keep locked down. These things often get filled with comments from bots and spammers or worse . . . salespeople (yes . . . you.) Let me know what antipattern you see or run across. The ones I listed are the most obvious to me, but it's not a comprehensive list. Post your example in the comments and make me say ‘Oh yeah!’ or ‘hmmmm . . .’
Recommended Reading/Listening
I mentioned that I recorded a podcast with Aled Sage and Alasdair Hodge who work for Cloudsoft. I love talking to people but I don’t generally just agree to go on random podcasts with people I haven’t met or talked to before. However, I saw they interviewed Ashley Hromatko and they immediately gained credibility with me.
Ashley is the Sr. FInOps Manager at Pearson and is a leader in our profession. When she talks, you know she has good content that is based in real-world application of FinOps best practices.
Here is her interview on Cloud With A Chance of Brain.
Playlist
I’m going to try to stick to the theme of people from Scotland with my playlist. The Scots are terrible singers. Great yellers! Bad singers. The Proclaimers are their best known musical export. And following the brand, The Proclaimers are great yellers and if you are really kind and want to like it, their singing is . . . on par with the typical UK Eurovision result.
I’m Gonna Be (500 Miles) - a song about how stubborn the Scottish can be. No need for any other form of transportation. They are gonna walk those 500 miles. Car? No thank you, and they will walk 500 more just to spite you. Just to be the man who made you wait the time it takes to walk 1,000 miles and show up at your door all winded.
There’s a Touch - I actually really like this song. I can’t even tease it.
Throw the R Away - if you ask a Scot to enunciate, you take away their slurring superpower.
I usually try to come up with 5 songs for a playlist, but as I said . . . not great singers. Three is plenty.
As an ultra-distance runner - going 100 or even 200 miles at times - I am really starting to be inspired by your music selection now. Some valid FinOps observations in this one as well.