An article published in The Register states that object storage has missed its opportunity to serve the secondary storage market. How did this happen?

Let's rewind the clock a bit. A few years ago, object storage was the new technology that promised to replace file system in the secondary storage market. It would scale beyond the limits of file systems and deliver better costs. The trade-off would be lower ease-of-use with new protocols that required rewriting of the applications. File systems would keep its place as the tier-one storage solution, because of its superior performance. The world order was clear with object storage elbowing file system out of secondary storage and gradually squeezing it out of tier one as its performance improved. The marketing message was that the role of the file system was rapidly diminishing, because its technology could not keep up with the data explosion.

What went wrong? Why are pundits writing articles about the missed opportunity of object storage?

First of all, let's look at the market. The storage market has few prospects that reach the hundreds of petabytes with many more in the range of a few petabytes. So scalability is an important argument, but it's difficult to make money at the edges of the distribution serving limited opportunities. The customers also want solutions that are easy to implement and object storage has not delivered the multi-protocols solution required by the users.

On the technology side, file systems have evolved quickly plugging the opening that was left open for object to grasp. Clearly file system starts with an inherent advantage over object, because it's ingrained in the data center and has very low adoption costs. Also file system capacity was not the weakness expected with disk size reaching 10 terabytes today and the core of the market in the petabyte range. File system continues to address the capacity requirements of most users.

What remains is the cost argument. There again the advantage of object seems to be evaporating. Erasure coding was a key differentiator of object by delivering the availability and reliability of triple replication with low overhead (50 percent or less). New file systems have caught up thanks to the introduction of high-performance erasure coding. With this new technology, file systems offer the same low overhead without sacrificing performance.

In this new world order, object storage is becoming a feature or back end of the file system. I think it would be fair to say - and I talked to users who have agreed -  that modern file system offerings are delivering the scalability, performance and affordability that most customers need these days.