Scality Looks to Remake Cloud Storage With 'Organic' Approach
Have the SAN and NAS systems of storage run their course?
At a time when unstructured data are proliferating in the cloud, startup vendor Scality is banking on a fundamental shift in the industry that will usher in the rise of object-based storage.
"We think that storage is about to be virtualized on commodity hardware. This major transition will happen slowly, over a 10-year period, but it will happen," said Scality founder and CEO Jérôme Lecat.
"It is fueled by two trends: The move to the cloud moves storage needs towards service providers, and the gigantic growth in the demand for storing unstructured content."
Scality's answer is to offer multiple storage nodes running on basic commodity x86 servers, creating a dynamic, self-healing network that flexes and scales to provide users with hundreds of petabytes of storage.
As for the company's novel approach to storage, which it dubs "organic," Scality explained its model as follows:
Organic storage demonstrates many of the features of a living organism: The storage is spread among a collection of nodes (like cells), each made of generic x86 servers with direct attached storage. These nodes share a highly parallel distributed intelligence which has no single point of failure does not depend on any single component. As a result the system is resilient, self-healing, adaptive, location aware and constantly renewing.
Lecat noted that Scality is not anticipating that SAN and NAS storage approaches will fall completely out of use in the near term, but rather that the company's object-oriented system will emerge as a viable alternative, particularly at a time when end users are increasingly expecting to store emails, music, photos and other content on the Web.
Scality's original story traces back to 2005 when Lecat, then the CEO of a firm called Bizanga, attended an industry conference where Mike O'Reirdan, a distinguished engineer at Comcast, spoke of the rising demand for Web-based storage. Three years later, in search of a new product for Bizanga, Lecat turned to the large Webmail providers who made up his customer base, and they expressed their interest in new storage technology for email. Scality was spun out as an independent company from that development effort.
"As a result, our storage technology is designed to handle the high-performance load of millions of users accessing their emails," Lecat said. "It is a great heritage."
In March of this year, Scality snagged $7 million in series b venture funding in a round led by Idinvest Partners. Earlier this month, the firm refreshed its RING organic storage product, promising "a new class of object-based storage that remains fresh and resilient for decades."
Lecat explained that the biggest challenge Scality faces is illustrating for customers the company's dramatic new approach to storage in the cloud.
"Most people don't know about 'object storage,'" he said. "For this reason, we actually describe our technology as 'organic storage,' because it conveys the right concept: storage is distributed over a large number of servers, which can individually be replaced, exactly like living organs are made of cells which constantly get replaced by newer cells."
Lecat also pointed out that established storage vendors aggressively promote their products' ability to handle enormous quantities of unstructured data, even when Scality is convinced that its object-oriented approach is superior.
"We invite potential customers to trial our software, and they usually realize only then how much easier it is," he said.
Looking ahead, Scality is hanging its hopes on the coupling of storage and processing power to handle large data stockpiles.
"We believe that for massive amounts of information, it is very important that storage and computing be located on the same systems," Lecat said. "This can be done by bringing storage cache to the computing machine and also by bringing the processing in the storage. SSD vendors pursue the first one, and it is great when a large cache is enough. We will bring about the latter, which is the best solution for search loads within big data."