The Case for Ephemeral Environments
In recent months, I’ve heard about ephemeral environments from many of my engineering friends. As an engineer by training myself, I decided to dig further into the market to see what the potential is.
What is an Ephemeral Environment?
Ephemeral environments are on demand and temporary environments that allow software teams to run tests on new features and collaborate asynchronously and in parallel with product, quality assurance, and sales teams. It also allows DevOps (developer operations) team members to not have to act as reliability engineers for all environments (eg. local, development, staging, production).
How does the Software Process Look Today?
The typical lifecycle of a software team is to develop, test, integrate, push to staging and push to production. Software is getting increasingly complex and the recent trend is to run containers with the latest code and dependencies. However, the only area that is not containerized yet is the development stage. Engineers develop on local machines that do not reflect the production environment and often run into issues when their code is integrated into a shared staging environment. There has been a movement called CI/CD in software development where automation is introduced and new code is continuously integrated, continuously delivered, and continuously deployed.
However these static shared environments are still inconsistent, and costly to maintain. This is because they are often pay-per-hour machine or virtual machine instances and are usually only used during work hours.
Why are Ephemeral Environments Important?
The reason ephemeral environments are increasingly important is that it helps teams break things even earlier in the cycle to prevent blocking each other further down the process. A shareable ephemeral environment with replicated data can be automatically spun up and sunsetted on each feature change which aids rapid testing and parallel processing. The beauty is that ephemeral environments can theoretically work for any company size, of any maturity, and with any workflow.
This is even more significant for software developers on smaller teams because they do not have a dedicated in-house DevOps team. Therefore they end up having to maintain internal tooling, debug pipelines, and set up development environments themselves, taking time away from development.
The beauty is that ephemeral environments can theoretically work for any company size, of any maturity, and with any workflow.
Ephemeral environments sit firmly within the DevOps industry. But it also sits within the Developer Tools and Software Testing industry because the real end customers are the developers and QA engineers, not the DevOps team. This is because ephemeral environments are meant to be spun up and down seamlessly without the DevOps team being involved on a daily basis. They most closely mimic end-to-end system testing because it involves testing the complete application environment that mimics real-world usage.
There are several approaches that we can use to get to an estimated market size for ephemeral environments. When I’m thinking about market sizing I’m looking at the use cases and validating the demand.
Industry Average Deal Sizes
I’ll first break down the different potential deal sizes. There are different EaaS pricing models such as charging by team, by user, or by specific resources needed. For this market sizing, I will look at the per-user per month model.
From industry averages and speaking with my network, an average team of engineers is around 8 people. But the reason I am not using the team business model here is that it’s hard to estimate how many software teams are in larger companies.
- $0 / user / month for companies just starting (0–10)
- $20 / user / month for small size companies (11–100)
- $40 / user / month for medium size companies (101–250)
- $100 / user / month for large size companies (250+)
I do not foresee companies willing to pay a subscription of more than $100 / user / month for strictly ephemeral environments (although money will be saved in other areas). As companies add more complimentary features, the price point can go up.
- End User Approach
According to the US Bureau of Labor Statistics 2021, here is the breakdown of roles that are relevant:
- Sales engineers 41,000
- Software quality assurance analysts and testers 74,000
- Web developers 81,000
- Software developers 1,932,000
- Computer programmers 444,000
When I think of a customer base, I think of the above roles. Because all of these roles can benefit from using ephemeral environments in their daily workflow, I will add them up. This adds up to 2.5M but I will round up to 3M as the number may not capture edge cases and factor in the growth of software teams in just a year.
The total market size from the table above is 2.02B.
2. Research Approach
In this approach, I would conduct extensive research into the biggest providers of DevOps environments (eg. VMWare, IBM, RedHat, Chef, AWS) and get usage data to figure out their active base or how much money they made (10K or 10Q if available). Some companies may break it down by sector and talk about active users and growth. Sometimes this would be discussed in the investor relations calls as well. Then I would speak with 10–20 experts in the field from my network or platforms such as GLG to further understand the market.
With the recent work from shift/hybrid work model shift, companies are searching for tools to help their employees do their work more efficiently. Ephemeral environments are starting to make a movement in the market and can soon overtake traditional continuous integration (CI) platforms as one of the most valuable assets for engineers.
According to studies, DevOps practices increased 57%; from 26% in 2017 to 41% in 2020 and is projected to continue to grow at a fast pace. Docker also published their usage numbers in the last few years. There are now 7M active Docker Hub users, up from 5M a year ago.
To fully understand the growth in the ephemeral environment space, a still nascent concept, I would get expert opinions to learn what the market was like in 2020, what it was in 2021, how they think it will be in 2022 and several years down the line. I would ask them questions such as: What existing tools do you see them replacing? Do you see this fitting in nicely with the automated testing market? It won’t be surprising to see the market grow at over 10% CAGR in the next 5–10 years.
Competitive Landscape (competitors + their positioning)
Kubernetes is an open source system for operating containerized applications at scale and is the industry standard. Kubernetes itself has documentation on how to create ephemeral containers but requires a lot of technical know-how to do so.
There are many container solutions in the market that are part of the Docker ecosystem such as: Compose and Swarm. Docker is so popular that it has a 83% market share in the containerization market.
Today, a huge proportion of companies that aren’t legacy enterprises use some combination of Kubernetes and Docker to containerize and deploy their software. So in that sense, the container competitive landscape is fairly concentrated.
However, in recent years, many companies have started to build SaaS models on top of systems such as Kubernetes that offer companies more testing capabilities and specific feature sets. These range from end to end testing platforms to automated testing platforms to cross browser testing platforms. As a result, the market is still fairly fragmented with many different players that are emerging, which I will cover below.
According to studies, DevOps practices increased 57%; from 26% in 2017 to 41% in 2020 and is projected to continue to grow at a fast pace.
Companies such as Amazon, Microsoft, Google, IBM, and VMWare all have their own solutions to deploy, test, and scale containers that are built on top of Kubernetes.
However, none of these companies provide customers with the option to easily create ephemeral environments that reflect production. These legacy incumbents are expensive to maintain, and require a lot of training to run. They are more targeted towards enterprises that have high requirements and take months to onboard onto a platform that suits their many needs.
Emerging companies are increasingly building off established platforms such as Docker and Kubernetes. Successful companies that preach CI/CD include CircleCI and Chef, and offer a full suite of solutions that include hosting and infrastructure management.
There are new open source projects such as HashiCorp Vagrant, and ephemeral.run (created by OpenGov and Infracloud) that are built on top of Docker and GitHub actions, and helps developers test seamlessly within GitHub.
Companies within this space
I’ve been following 8 pre-seed to series A startups within this space that are helping companies create ephemeral environments faster and cheaper: Shipyard, Roost.ai, Release, Garden, Bunnyshell, Loft, webapp.io, and Signadot. Many of them aim to educate the market about why ephemeral environments are the “new kids on the block” and I will look to disucss them further in a separate post.
The ephemeral environment market is still nascent. From speaking with my network, many people are slightly hesitant about how ephemeral environments work at scale, but see the value. However, compared to legacy incumbent platforms such as Amazon EKS clusters, ephemeral environments are less complicated and more intuitive.
As the software industry starts to understand the benefits of ephemeral environments, I believe there will be increased adoption in the next few years. My hypothesis is that ephemeral environments will gain popularity in SMEs initially and later expand to enterprises. There is a great opportunity to have companies merge their CI/CD with ephemeral environments to form a unified CI/CD and review process for every commit. It will ultimately be a good way for enterprises to optimize large capital investments so that services are not running on idle on hardware.
My hypothesis is that ephemeral environments will gain popularity in SMEs initially and later expand to enterprises.
Ephemeral environments fit venture firms who have developed theses in areas such as cloud infrastructure. Here are some of the reasons:
- While not the biggest market, the market for ephemeral environments is over $2B and will continue growing at a good CAGR
- There is good defensibility because this is not an easy problem to solve. There are many software teams in the US and globally that have specific workflows that they have developed or are developing (albeit generally for the same few tools). If a company is able to develop a solution that fits with different processes, they can build a competitive moat and land and expand with an existing customer
- Being that this is a nascent and still fragmented market, there is opportunity for startups to build differentiated technology
- Entrenched incumbents do not offer quick set ups for ephemeral environments
- This will save companies a lot of money in the long run because they don’t have to constantly think about how many machines they need to run end to end tests
- Almost most importantly, the startups in the space that focus on ephemeral environments have already began to validate the market with the customers they have been acquiring
But there are some risk associated:
- Large teams will provide higher ACVs to emerging companies. However, many large software teams have automated test systems that have taken years to build and are built off open source community projects such as Buildbot and may therefore be less willing to change their workflows
- The new Environment as a Service (EaaS) business model is so new that companies that are already paying for other subscription services may be unwilling to go for another unless there is clear immediate ROI
- There are many free open source projects that offer sophisticated teams an opportunity to build their own infrastructure without leveraging the technology of startups that cost money
https://www.slintel.com/tech/containerization/ https://devops.com/a-new-devops-test-environment-platform-is-needed/ https://www.datadoghq.com/docker-adoption/ https://www.docker.com/blog/docker-index-dramatic-growth-in-docker-usage-affirms-the -continued-rising-power-of-developers