An object-storage-as-a-service startup has hit unicorn status, touting a valuation of more than $1 billion.
MinIO Inc., founded in 2014 and located in Palo Alto, Calif., began as an AWS S3-compatible object storage vendor but quickly moved beyond AWS to become compatible with every public cloud, according to Jonathan Symonds, CMO at MinIO. Focused "obsessively on simplicity," MinIO has continued to add new features and build out the product since it was founded eight years ago, he said.
In this Q&A, Symonds and MinIO CEO and co-founder Anand Babu Periasamy discuss the direction the startup is headed as it comes off its Series B funding of $103 million, which brought its total funding to $126 million and a valuation of more than $1 billion. Although they didn't share specific plans, the executives said they intend to invest in marketing and education, given that object storage often causes confusion.
"I would say the area of focus now is going to go toward content creation," Periasamy said. "There's so much confusion. And it's our job to go educate the industry."
With the latest round of funding, what specific areas is MinIO going to focus on?
Jonathan Symonds: MinIO has grown up by looking at developers. Developers are the engine of value creation in the enterprise today. We're going to continue to serve them in the best way possible. The best way possible is to pour money into education, learning and development, certification, technical marketing, the product, making [MinIO] easier to adopt, making sure that people understand why object storage is operating as primary storage -- all the things that you can do when you have performance at scale. Telling that story, that's going to be what we invest in.
But MinIO doesn't look like a normal organization where we're going to go hire a huge customer success team. Customer success is a software problem to us. We're going to invest in it from a software perspective. We're going to solve these problems with code: We digitized our sales channel. We digitized our customer support and functionality. We're going to continue to do all of these things, but that doesn't require us hiring giant teams of people and opening up branch offices.
How does MinIO differentiate itself from other object storage vendors?
Symonds: Data is increasingly generated at the edge, and MinIO is uniquely qualified [to tackle that] for two primary reasons. One is that object storage, which is the only thing that MinIO does and the only thing we've focused on since our inception, is now the primary storage class for the cloud and is the primary storage class for Kubernetes. … It's also Kubernetes native. As Kubernetes takes over the compute landscape, object storage has taken over the storage landscape, which leads to a requirement for multi-cloud.
MinIO … runs on every single public cloud. We have almost a million deployments within the public cloud. It runs on every single Kubernetes distribution. … MinIO allows the enterprise to turn the public cloud or any other cloud into a commodity infrastructure of cheap CPU, network and drives. That provides a significant change in the way that you approach your storage requirements because you have a consistent interface from the edge all the way to the core.
You mentioned multi-cloud and Kubernetes. How did you plan for those trends?
Symonds: From the very beginning, MinIO recognized data would exist outside of AWS, and that was going to require supporting data on multiple platforms. … We did not fully appreciate the speed with which Kubernetes was going to remake the industry as the vehicle by which you could now put your code almost anywhere as long as you were truly cloud native.
The other thing that occurred is that multi-cloud emerged in the enterprise, almost by accident. It's a strategy now, a business continuity strategy, as we've seen from Amazon. If you don't have a multi-cloud backup plan and your entire business runs on AWS, you're in trouble. We saw enterprises finding this: They may be multi-cloud because they like the natural language processing capabilities in Azure, or they like some other machine learning vision capability from Google and they like the other database services they get from AWS. They became multi-cloud, whether by intent or by accident. What we found is that they needed a consistent data infrastructure across all of those [clouds].
Anand Babu Periasamy: In the long run, public cloud, private cloud -- everything will look like AWS. I won't discount the fact that Amazon is capable of becoming the world's data infrastructure. Our bet is just against. … Our bet is that AWS will be one of many players, and the rest, they are not going to be SAN, NAS, HDFS. … We will make everything look like AWS.
What's the growth potential for an object storage vendor?
Periasamy: Our customers are producing new data that is more than all of the historic data put together. It's a consistent pattern across customers. [They] started with 500 TB and now they are talking about 6 PB, 10 petabytes. For us, the market is growing fast.
You will see companies like MinIO having accelerated growth while many storage companies are dying. That's just the nature of this market. We knew when we entered that the market would consolidate -- this is a winner-take-all market. And it may be not one winner, maybe two or three. But the second and third [place winners] … [are] going to pick up the leftovers. You will see [this play out] in the next two years.
Symonds: Object storage is going to continue to grow for two primary reasons: One is the fact that the S3 API … is the modern API for which all applications are written to today. … You see that in the success in AWS S3. Two, for fast object storage, modern object storage, like MinIO … every workload now comes into play. Workloads that were the bread and butter of standard NAS or databases in the past are now almost entirely focused on using object storage as their primary storage class.
What's the benefit of providing object storage as a service?
Symonds: In the past, doing something as ambitious as object storage as a service was beyond the capability of somebody who was not highly trained in [object storage]. Now, people who are storage admins who can't spell Kubernetes can now deploy [object as a service]. The reason they can do that is usability. They can manage giant, petabyte-, exabyte-scale infrastructure with a single person.
Will larger SAN/NAS hardware vendors like IBM, Hitachi, Dell EMC and NetApp continue to dominate the market or will new players like MinIO break into the top spots?
Periasamy: While object storage is growing, the SAN/NAS players, NAS in particular, is getting beaten badly because modern applications are not speaking NAS. The current market is so big that, even on a slow decline, it is still a big enough market. One thing we are seeing in file and block, the block market itself is changing. The SAN vendors are now talking about disaggregated storage, the complete reversal of the [hyper-converged infrastructure] story. They're talking about disaggregated because that's the theme of the cloud. In the cloud, when you talk about disaggregated storage, it means [something] completely different. [When] vendors talk about disaggregation between the application code as stateless services … And after it's processed, where does the data persist? Object store. Even when they write to databases, the databases themselves … modify their back end to go object storage.
There are legacy applications that depend on block storage and NVMe over fabrics will become more popular there in the coming years. But the overall block storage market is on the decline. And a simple way to do that is to look at AWS, or any one of the clouds, what percentage of the data is in block versus file versus object? You will see that file and block are miniscule -- file is almost nonexistent.
Symonds: What we've been doing over the last year, we have been building out this concept of true multi-cloud. A lot of people say, 'I do multi-cloud in this one way or the other, but it requires unnatural acts to get there.' In the past two weeks, we've launched click-to-deploy [capabilities] on AWS and Azure. … Every cloud worth being on, we have gone out and invested in heavily to make it click-to-deploy simple on each one of those clouds. We're reaching the end of that kind of cycle at this point.
Next, we are going to turn our attention to things having to do with running giant infrastructure at scale, through our subnet interface to our customer success portal. … We're going to turn our attention to some of these next higher-level problems that object storage is uniquely qualified to solve because of the scale that you must contend with.