by Felix Xavier on September 19, 2012
Storage industry went through a lot of technology advancements in the late nineties when NAS and SAN got popular. This was followed by a relatively quiet period for over a decade, marked with incremental innovations such as logical volumes and deduplications. The wave turned again, with the high-profile acquisitions of 3PAR and Data Domain, which sparked off a new interest in storage technology. A plethora of startups came up in the last 2-3 years, with the best minds once again focusing on storage technology. On a high level, these startups can be categorized into broad groups, based on their storage vision.
The first category attempts to reduce the importance of storage devices, by baking the current storage intelligence right into the app server. Technology advancements like SSDs and common availability of 10GigE also aid in the reasoning (to an extent) that there is no real need for separate storage devices. The interface between disk shelves and storage controller has already moved from FC (fiber channel) to SAS. I expect this to shift soon to IP for obvious reasons. So, this group questions the very existence of storage controllers with the following argument – “why don’t we just leverage IP-based disk shelves and bake storage intelligence right into the app server?” Taking this further, even the huge disk arrays can probably be eliminated, as internal SSDs will soon suffice to deliver the desired storage performance.
The second category is all about making the current storage a lot more intelligent, sophisticated, and grid-like, with the storage controllers set to solve the new class of problems in delivering predictable storage performance, data mobility, and high speed storage specific processing etc. None of the intelligence will be attached to physical resources, but will rather be dynamic and software-defined, including servers, switches, firewalls, and storage. Here, storage is expected to become the center of the infrastructure, enabling data to be available at the right place, within the right time. Applications can then just spawn a compute instance on top of this data.
Companies like Google, Facebook, and Amazon spent a lot of time and resources in building infrastructure to cater to their scaling demands, implying any new company to spend similar time and resources to effectively compete with them. However, with this new class of storage controllers that the second group is trying to build, any new company can get access to similar infrastructure almost instantly. Hence, with this second group’s vision, the competition will be between two applications, rather than between two infrastructures. Obviously, this freedom from infrastructure hassles is great for both evolution in technology and delivering better products to enterprises and consumers.
I strongly believe in the vision of the second group and needless to say, the company I helped found, CloudByte, is gearing up to be at the forefront of this change.
Founder & CTO