Startup offers fixed-content archive for multiple applications
Former CTO of the New York Times has started Archivas, a company that archives fixed content from different applications in a single, centralized repository for half the price of EMC's Centera.
Archivas Inc. has emerged from stealth mode today with a product it claims fills a hole in the storage market for the single, fixed-content repository that can be shared among multiple applications.
Andres Rodriguez, founder and CEO of Waltham, Mass.-based Archivas, started the company to solve a very specific problem. He was CTO of the New York Times and was faced with the daunting task of digitally archiving over 150 years worth of back issues.
Rodriguez found there were no tools available to help him do this in an efficient way. The best he was offered was separate NAS systems for each application that he needed to archive. Using distributed dedicated servers, he set up the best system he could to get all the data online, but he says the management of distributed servers with separate islands of NAS was inherently expensive.
In May 2003, Rodriquez quit the Times to start Archivas. With a team of 20 engineers, he built ArC, an object-based file system designed specifically to store fixed content, such as medical images, voice recordings and satellite images in a single, centralized repository.
Archivas has since received $6 million in Series A venture funding and a $100,000 grant from NASA's Small Business Innovation Research Program. The Goddard Space Flight Center at NASA has installed a beta version of the product to explore its capabilities. Researchers at Goddard are looking at ways to consolidate the storage of thousands of satellite images which currently live on direct-attached storage.
The Archivas software runs on industry standard servers and right now has been deployed in a maximum configuration of 10 nodes, for about 10 TB of storage. The company says it can cluster hundreds of servers together, although this has yet to be even attempted, let alone proven. It's also worth noting that ArC can only withstand two simultaneous failures. In other words, if two servers die, the whole thing goes down. "Three of four failures would definitely be problematic; you would need a remote replication scheme," says Asim Zaheer, VP of marketing at Archivas. [After going to press, Archivas got back to us and said it does have a scheme for failover.]
Downtime fears aside, Archivas' object-based management approach offers some interesting features. It stores files, meta data (information about the file) and file policies together. As a result, ArC stores and retrieves archive objects—not volumes or files—so that users can change applications or system platforms without impacting their ability to store and retrieve the content stored in the archives.
And with an eye toward regulatory compliance, the company has created file retention policies, content authentication, data shredding and various levels of security for sensitive data.
Archivas faces steep competition from EMC Corp.'s Centera fixed-content system and there are other small companies such as Isilon Systems Inc. and Permabit Inc. that are also gunning for this space.
ArC is slated to be available in September, priced at around one cent per megabyte, or about half the price of EMC's Centera, the company claims.