Minio, for a simple object storage

Minio (www.minio.io), recent object storage player and good surprise of the last IT Press Tour last week, has a simple mission: provide a simple and powerful open source object storage.

The first interesting thing about the Minio is the DNA of the company as the company was founded by Anand Babu Periasamy, famous in the industry with the Gluster story. And we all remember that Gluster was one of the first approach doing what people name today Software-Defined Storage.

In fact, the software did a pretty good job aggregating independent servers in a shared nothing mode and offer a large storage pool. The product was a success being adopted by many famous companies and organizations of all type. What an impact for an open source project, good lessons for everyone. And the effect of a rapid success is often an exit, we hope a good one in that case (it was) with Red Hat for $138M in 2011. With that acquisition, Red Hat has confirmed his deep interest in storage after Sistina in 2003 for $31M and later strongly pushed with Inktank acquisition for $175M.

Now with Minio, the story has some similarities but the context and the market climate is also very different. AB is convinced that the next era is the Cloud Storage and we even already entered in it. For that, the key assumption is that Amazon S3 is everything as Amazon has made a great job promoting and evangelizing the access API. So clearly in the future, we should have in majority something linked or relate to S3 so starting an object storage product today, you must provide and support this access API, no doubt, the reverse would be a big mistake. Minio is written in GO and uses an Apache License v2, the product was designed to support immutable data which today represents the vast majority of the data stored on the planet.

In term of product Minio has 3 components: the Minio server, the Minio client and a SDK but 2 flavors of the product will exist: Minio and Minio XL like the T-shirt. Minio supports only SSD and HDD and Minio XL provides distributed Erasure Coding (8 data chunks + 8 parity chunks) and the team made some developments to be deployed with Docker. Among all these offerings everything is ready except the XL flavor that should be available around summer 2016.

Let's wait a few months to see if developments will be aligned with the wish...