Today marks a special anniversary for MinIO. It has been five years since we had our first GA release on GitHub. The journey to that point and from that point (5,396 commits later) has been nothing short of amazing and we wanted to take a quick moment to recap our thinking when we started and where we are now.
Back in late 2014 when AB, Harsha and I were thinking of doing a startup, we went through multiple ideas, but every one led us to the same problem: that data was going to overwhelm older technologies like SAN and NAS. Right around that time, AWS had convinced the world that S3 object storage was the right answer for unstructured data and semi structured data - particularly at scale. We knew AWS can not possibly be the world's data center so we wanted to be AWS S3 for the rest of the world.
It was ambitious to say the least. Yet fast forward five years from our first GA release to where we are today 500M+ Docker pulls and a presence in more than 58% of the Fortune 500, and you start to get a sense of the exceptional growth and love we have from our community and customers alike.
Because we started in 2014, it gave us an opportunity to build the object storage stack to natively address the cloud native workloads. Thanks to the massive adoption of the S3 API now we have a huge ecosystem of applications which work natively, without a single line of code change, in MinIO from big data applications like Spark, Presto to AI/ ML like Tensor flow, Jupyter notebook, Pytorch etc to traditional back up applications like Rubrik, Cohesity, Veeam, Commvault etc. Applications like Kubeflow bundle MinIO natively as part of their stack.
One of the biggest reasons for the widespread adoption and huge application ecosystem is that we have been focused in our approach, we do one thing and one thing better than the rest which is S3 compatible object storage. Supporting NFS or Block storage or even Swift API natively would be like driving with a rear view mirror!
While our “one thing” has provided us focus it still requires discipline and principles. There are three that we return to time and time again:
First, be cloud-native in everything you do. We did our first GitHub commit in 2015, we GA’d our code on this day in 2016. We were literally born into the cloud. It was early days to be sure, but there was no mistaking what was happening. Because of that timing all we have ever known is restful APIs, microservices, containers, orchestration and automation. They are simply part of MinIO’s DNA. We were part of that ecosystem and that ecosystem is part of us. It seems that barely a day passes without some really cool, new, cloud-native use of MinIO coming to light. Our Twitter feed overfloweth with amazing content and these companies as well as those individuals become part of our every growing family. It even extends to the giants in the industry. Who does VMware, Veeam and Splunk use for their object storage demos and documentation? MinIO. Why? Because we are open source? That’s helpful to be sure - but the real reason is that we work with everything cloud-native. Right out of the box.
The second principle is performance. Everything we do at MinIO is viewed through the lens of performance. SIMD acceleration, encryption, erasure coding, metadata databases. We squeeze performance out of everything we do. The payoff is in our published numbers. READ/WRITE speeds of 183 GB/s and 171 GB/s on a 32 node NVMe cluster. 11/GB/s and 9 GB/s on a 16 node HDD cluster. Speed expands the application map. We looked at AWS and saw a machine that could do far more than simple archival workloads. We knew that to compete we needed to offer similar performance - across a heterogenous set of hardware. It was a massive challenge but we succeeded and continue to marvel at the various posts and tutorials about MinIO running on everything from drones and Raspberry Pi’s, to the latest NVIDIA chips. This in turn puts on the broadest set of applications from AI/ML, analytics, application workloads and yes, even archival.
The last principle is perhaps the most underappreciated - it is simplicity. Simplicity is hard. It requires discipline and commitment. It is the opposite of weak - simple is powerful and more importantly simple scales. Since our inception, MinIO has worked tirelessly to ensure everything we do is simple and this principle, as much as any other has allowed us to grow to where we are today - with more than 500,000,000 Docker pulls, 26,400 GitHub stars and more than 600 people contributing to our work.
So here we are on our fifth anniversary of our first release and we couldn’t be happier about the state of MinIO. It is a very exciting time for us, our customers and community are the center of our universe. Nothing makes my day like when I hear ‘MinIO just works’ .
In five short years we have become widely recognized as the most innovative object storage company in the industry. Our Kubernetes work is allowing us to fulfill our dreams of becoming a major player, perhaps the major player, in hybrid cloud storage. Our goals for the next five years are incredibly ambitious but we can’t wait to take you along with us on the journey.