One of the most interesting meetings I had this week (during an IT press tour of Silicon Valley) was with Minio. I also heard from the founder of the company, AB Periasamy, who has some really interesting ideas, and outlined his vision of “object storage for everyone”.
Object storage for the rest of us
Minio is working on µServer, an object storage server which is particularly small, efficient and can run in a single VM (or VPS). (It's actually working on two, but let me explain the basics first.) In addition, there are a generic S3 client and an SDK that will complete the picture.
Taking for granted Minio's promise that it has one of the best S3 compatibilities in town, the idea is to enable any user and developer to deploy object storage at any scale (starting with just a few GBs). The products are open source and as easy to use as a MySQL server.
We are not talking about the most advanced object implementation on Earth and maybe not the fastest or the most reliable but, in practice, I think the analogy fits perfectly: S3==SQL and Minio µServer == MySQL. As MySQL could be considered the SQL engine of choice when you need a good DB for your personal or SMB needs, Minio µServer could become something similar for S3 and objects.
Minio could be the answer for a developer who wants to build applications with freedom of choice, at any scale and on any cloud when it comes to object storage. It also means that end users won’t have any barriers when it comes to implementing object storage even for the smallest of deployments.
More than a micro server
Minio doesn’t want to be just a MySQL; this is part of a larger plan. In fact, the firm says an “XL” version of Minio is already in development and will have features like global erasure coding, for example, that will be aligned with the expectations of those implementing large scale object stores. Also this product will be 100 per cent open source, and the full featured version will be available for free.
Closing the circle
The strategy of this small startup is very aggressive and, even though the business model is all but clear, there is huge potential in what it is doing.
Developers like the S3 protocol. It’s easy to use and completely aligned to their way of thinking. Giving them an option to have a simple object store running in their laptops, a Raspberry Pi, a VPS or whatever they like, is something that will open up many different opportunities to write standard code based on de facto standards and give customers/users an end-to-end solution.
At the same time, if the XL version is good enough, Minio will be positioning itself as one of the most competitive object storage solution providers covering every type of scenario from 1GB to several petabytes or more. It’s too soon to say whether or not it will be successful, but one thing is certain… its ideas are definitely exciting.
Object storage is one of the hottest segments in the storage market at the moment, with two interesting things happening:
1. The number of startups in object storage is continuing to grow. This week I also met OpenIO and, even though the strategy is different, it has some interesting ideas and an open source core. (I’ll follow up soon on Open.IO after a more technical briefing about their technology.)
2. S3 is increasingly being considered a generic storage protocol and it is no longer associated only with very large distributed infrastructures. Most of the companies I met this week have an S3 interface or they are planning to have it soon… even Mangstor (a startup working on an NVMe fabric AFA) is experimenting along the S3 track.