If you are reading this, chances are you have seen the reference architectures below. Whether you are on AWS, GCP, Azure, on prem or a mix thereof, these have now proven to be reliable blueprints for building an Internal Developer Platform (IDP) that is truly enterprise ready.
You will have also noticed that the centerpiece of these architectures is the Platform Orchestrator. This is the de facto industry standard solution to make your IDP enterprise-grade. Thoughtworks added it to their latest tech radar, the CNCF PlatformWG discussed it a KubeCon NA 2023 and Microsoft highlighting it too.
Platform Orchestrator: what makes your IDP enterprise-ready
That’s why you need a Platform Orchestrator as the core of your enterprise-grade IDP. This is the configuration engine that orchestrates your entire platform layer. It integrates with all the tech and tools in your setup and turns a disjointed toolchain into golden paths for your developers.
Engineers can add any resource they need, spin up environments, roll back and much more, without having to wait on Ops teams. All following clear baselines that let you drive standardization by design, across all teams and workflows.
Operations teams no longer have to fight ticketops and manual configuration (and human error) drops by 95%. Check our product page for more details.
If you are building a platform for the enterprise, you have to look at this orchestration layer. So the question is, should you build or buy a Platform Orchestrator?
How you can think about it
When categories are young, there is a tendency to try and build everything yourself. We have seen this before with CI systems or container orchestration. Who today would imagine wanting to build their own Kubernetes?
With that, it’s probably not wrong to assume it’s not the first build vs buy decision you are evaluating. But just so we are on the same page, here are the key questions you want to answer.
- Strategic value: Is building a Platform Orchestrator an essential part of how your company creates value for your customers?
- Talent: Do you have the right people to build it? Do you have enough of them?
- Time to value: How long would it take you to build it yourself? And if you buy, how long would the implementation take?
- Cost: How much would it cost you to build it on your own? What would be the total cost of ownership (TCO)? What’s the ROI? And not to forget opportunity costs: could engineers spend their time on something else that is worth more than the investment?
- If you buy: How reliable is your vendor?
Below is an example build vs buy analysis one of our customers (financial institution, anonymized) did, you can download an empty version here.
Note: the build-vs-buy scorecard here is from Reforge’s Technical Strategy program.
This is obviously one of many possible examples. In order for you to have the necessary context for your own calculation, let’s break down the different questions.
Is it the core of your business?
Building an Internal Developer Platform might feel extremely important if you are in platform engineering. And it should. Therefore building the centerpiece of it, what makes your platform truly enterprise ready, might feel like the most strategic thing you can possibly do. The reality is, it’s not. The strategic value of your IDP is crucially dependent on how quickly and reliably you ship it so you can have a real impact on the engineering performance of your organization. Similar to how the underlying stack wouldn’t be a strategic priority (e.g. you wouldn’t start building your storage from scratch), building a Platform Orchestrator in-house isn’t either.
It’s important not to confuse the strategic value of having a Platform Orchestrator (extremely high) with the strategic value of spending millions on building and maintaining your own (arguable). If you are working for a financial institution, a healthcare company, or an e-commerce, building orchestration tooling is likely not the core of your business. Your value as a platform team is in adapting existing solutions (whether open source or commercial) to the specific needs of your organization (which only you can know), not in reinventing the entire wheel.
Do you have the talent to build and run it on your own?
Building a Platform Orchestrator requires more than a skilled platform engineering team experienced in K8s, IaC, Helm, and security tooling. Crucial to the success of the project is a seasoned product manager who has built similar systems, excels in user research, roadmap development, and ensures execution and delivery.
The architecture of the orchestrator must be adaptable to evolving technologies and unforeseen scenarios, such as mergers or acquisitions. The team should master key concepts like resource management, environment-specific variables, and RBAC design. Maintenance and ongoing development are essential, with clear SLAs and SLOs to ensure platform reliability and adoption, requiring a stable team committed for at least three years.
Estimating the cost of building a Platform Orchestrator is vital for securing top management sponsorship and showing a clear ROI case. Since we are using USD for our calculation, let’s take a US-based org as an example.
For an engineering org with 300 developers the calculation could look this this:
The table above shows a conservative estimate of what it’d take to build and maintain a Platform Orchestrator in house, similar in functionality to Humanitec’s. Note that $150,000 as estimated FTE salary is on the lower end for the US (according to latest Platform Engineering community salary survey US platform engineers earn approx 220k per year).
To keep the calculation simple, we assume it only takes one year to build the Platform Orchestrator (it’s likely more like 2+). In year one, an organization of 300 developers will on average employ 14 FTEs (including product management) to ship the Platform Orchestrator, that’s a total cost of approx. $2m. Just to maintain it and iterate on the features based on developers’ feedback, on average 6 FTEs are required. This means annual costs of approx. $900k.
So the total cost to build and maintain a Platform Orchestrator in house is around $5.7m. It’s important to note that this is still not the Total Cost of Ownership (TCO). Infrastructure costs such as servers, networking, storage, etc. necessary for development are not included in this calculation, nor expenses associated with ensuring security and compliance (e.g. certificates).
Finally, if you wanted the full picture, you’d also need to calculate potential efficiency and productivity tradeoffs, e.g.:
- Slower time to market (years instead of weeks if implementing a solution like Humanitec)
- Reduced developer efficiency (the in-house Platform Orchestrator won’t likely be as fine tuned on DevEx best practices as one like Humanitec’s, which is currently deployed across many teams and orgs of different sizes)
- Opportunity costs (resources and time spent building the Platform Orchestrator instead of other value-generating activities)
As well as risks and contingency costs:
- Risk of failure (this is not a rare occurrence, you should take the possibility of failure into consideration)
- Contingency costs (arising due to unforeseen challenges or delays, which is very often the case)
Reliability of the vendor
Yes, this is where we brag. But for good reasons. Humanitec is the de facto leader in the platform engineering space. We have been the main contributor to the community (whether platformenginering.org, the Slack or PlatformCon) since day 1. We have the most mature set of products, from the #1 workload specification in the space, to the Portal and of course the only truly enterprise-grade Platform Orchestrator on the market.
We have looked at thousands of platform setups across all industries and org sizes in the last 5 years, there is none that has as much data or knowledge on what it means to successfully build and ship an IDP. Leading consultancies like McKinsey&Co work closely with us on key initiatives like the reference architectures we launched at PlatformCon23.
Our enterprise customers (including Fortune 100) appreciate the flexibility and reliability of our Platform Orchestrator. No churn so far, NPS over 85. Security is to the highest standards.
We have the leading platform engineering experts on our team (built the IDP at Google and Microsoft, core contributors to the community) and in-house squad of professional services wizards that will go above and beyond (check out our ProServ page for more information).
Time to value in 14 days
We believe that most platform engineering initiatives fail because they move too slow and try to win over all stakeholders at the same time. The top performing teams start small and fast with a minimum viable platform that can demonstrate its value quickly, and then expand. This lets you deliver value in a timeline of weeks rather than years. That’s 2 weeks vs 2 years (and likely millions saved) to prove the value of your platform engineering initiative within your organization.
Your platform team can start immediately with our team license for only $999 per month after a 30 days free trial. Use the team version to get started ASAP and prove value to everyone quickly.
Then upgrade to the enterprise plan to unlock enterprise features (RBAC, SSO and more), and get the best expertise the industry has to offer by working with our ProServ team of platform architects to make sure you ship your target platform design fast and reliably.