HashiCorp Field CTO Weekly
Volume 108: Organizational Velocity, Cognitive Load and Platforms
Number One with a Bullet
Starting on a positive note, GitHub released their annual Octoverse report - The State of Open Source Software. HashiCorp Configuration Language (HCL) was the fastest growing language on GitHub with 56.1% growth YoY, no mean feat considering there are 500 primary languages used on the platform.
Per the report: “HCL saw significant growth in usage over the past year. This was driven by the growth in the popularity of the Terraform tool and #IaC practices to increasingly automate deployments.”
This tracks with what we’re seeing, as Terraform has been downloaded hundreds of millions of times, and we know platform teams from Barclays to General Motors and even our friends at #GitHub are just a few of those who are adopting Terraform as they standardize their approach to cloud infrastructure automation.
See the full Octoverse report here which is - as always - full of useful and interesting nuggets: https://lnkd.in/ehF_UK5f (I am pleased to see Java still holding strong after all these years).
Storm Clouds
Now the other side of the coin - what an insane time in tech!
Elon picks up Twitter, and causes no shortage of chaos in the process. (I personally chuckled at the parody verified accounts fiasco). FTX melts down from a $32 Billion USD valuation to imminent bankruptcy within a week’s time! Tech has seen a bloodbath in layoffs, and the three biggest cloud providers miss their growth targets. Despite all that cheery news, all is not lost.
The late great Formula 1 champion, Ayrton Senna once said: “You cannot overtake fifteen cars in sunny weather - but you can when it’s raining.”
Today, it’s raining worry, and concern - to which none of us are immune. Despite that, not all of us will let these headwinds stand in our way. Some of us will be busy overtaking those fifteen cars by overtaking competition and disrupting our industries.
Velocity, DevOps and Reality
How do we go fast? There’s a lot of advice out there on that topic. But it’s not necessarily as easy as the power-point slide describes. Forrest Brazeal performed several catchy tunes at DevOps Enterprise Summit on the topic (it’s really worth the watch). Many of us can agree that these “Digital Transformations” are messy.
The technical and infrastructural parts are certainly complex but in my observations over the past years most organizations were able to tackle them quite well. Investing in automation is key and you should make sure that this is understood by all stakeholders including infrastructure, security, networking and application development teams.
The modeling, organizational and cultural parts are the hardest ones to tackle. None of these are simple and none of these can be done quickly by bringing in some high profile experts: the team needs to do the work and have the patience.
IT Revolution Press’s latest journal has a great real-world case study about Fannie Mae, the US-government backed mortgage company, and their efforts to transform their business. The team focused on establishing a new operating model, with the goals of enabling faster delivery and greater efficiency.
The extent of role transformation should be great enough that a customer feature can be picked up by a team and delivered to production with minimal to no interaction with other teams.
Fannie Mae is pretty established - they have over 250 Agile teams, and they did all the right things, including creating agile squads of product owners, scrum masters, full stack developers and DevOps engineers. They focused on automation and discovered boundaries within which teams can make a high percentage of decisions without having to coordinate with others. Cross-team coordination will always slow you down, no matter what fancy tech you will use later on.
“[The efforts] yielded significant results in terms of a streamlined process, transparency, efficiency, and savings (both tangible and intangible). Along the way, we invested in improving the developer experience, such as enhancing the tools they use, and refining the developer feedback loop by automating and soft/hard gating the DevOps pipelines.”
Sounds great, right? According to Fannie Mae, the work is far from done. The developer teams are picking up SRE responsibilities and the company has adopted a “You Build It, You Operate It” policy. The policy has caused a lot of concern around cognitive load. This trend to shift left with security, operations and infrastructure leaves a lot of responsibilities that engineers need to manage; and less time to focus on things that provide differentiated value. If it’s not one thing, then it’s another.
Enter the Platform
Much as HashiCorp has been recommending, Fannie Mae is exploring internal development platforms in order to mitigate as much of the extraneous cognitive load as possible. The article goes into the value of funding and supporting these internal platforms. It’s not uncommon for enterprises to dedicate 20% of their budgets on building out these capabilities (Shopify considers it important enough to dedicate 50% of their budget on these types of endeavors).
By shifting from a shared service model to a platform, teams can now be the masters of their own success or failure. I really like that the article mention points that re-affirm much of what HashiCorp professes:
Be pragmatic balancing standardization with the need for flexibility
Make the standard path easier than ad-hoc solutions
Automate as much extraneous work as possible
Develop known good practices by default
Ensure for operational simplicity
Adopt an API-first approach for all platforms
Done properly, we nudge developers to “fall into the pit of success.” These defaults have power. People rely on defaults in unfamiliar environments or when they are unsure what to do.
We are not necessarily in the sunniest of economic landscapes, and that provides opportunities for disrupting your competitors. There’s a lot that can be done to ensure your teams can be successful and move quickly, one of those efforts you can adopt is a platform-mindset. Platforms can help reduce your cost, improve agility and efficiency, provide better visibility, governance and auditability; and help your teams scale in time. I personally look forward to seeing how this approach evolves in the industry over the next few years.
Until Next Time,
Ray