The deal was rumored last week but only officially announced yesterday, when a source close to the deal also confirmed the purchase price.

Under the brand name Mozy, Berkeley has only been on the market for two years. But has already won around 300,000 consumer customers and 8,000 business clients. Although almost all of its business customers are small companies, it has won a contract with manufacturing giant General Electric, for whom it will provide online backup of 300,000 laptops and PCs.

While EMC would not say what its plans are for the business, Berkeley said that the back-end and front-end technology that it has developed could easily be used to sell online storage services to EMC’s more traditional mid-sized and corporate business customers.

Berkeley attributes its rapid growth to its low prices, which it says are enabled by what it claims is a cutting-edge, home-grown storage grid installed in its two Utah data centers.

Presumably that grid will now help fulfill the promise that EMC CEO Joe Tucci’s made in April this year, when he said that the storage giant would soon launch online services, including but not necessarily restricted to backup.

The use of online backup services is already growing among small and mid-sized businesses, and among large corporations wanting to automate backups at their remote offices. According to a report issued by the AberdeenGroup [sic] last month, the number of businesses that are using such services to protect remote office data has doubled in the last twelve months. IBM, HP and Iron Mountain already offer online backup services, and Symantec was scheduled to begin doing so this summer.

Google has previously been rumored to be negotiating to buy Berkeley. Coates said Berkeley was visited last year by a well-known blue-chip business that is currently developing its online services, and is attempting to develop its own back-end storage grid for that purpose. They made some chatter and noise [about buying Berkeley] but nothing came of it, he said. They were the ones we took that GE deal from, he claimed.

The Mozy business-tailored service, which includes near-continuous data protection and bandwidth throttling, was only launched earlier this year, and has been taken up quickly. Nevertheless the huge majority of Berkeley’s customers are consumers.

With the exception of a relatively limited business selling Dantz-originated Retrospect backup software, EMC has never before been a supplier of consumer products or services. Tucci has said more than once that the company has developed consumer products but has held back from launching them because the company has no expertise in consumer marketing.

Berkeley said its business has mostly grown through word-of-mouth, and that it has done only limited advertising, Our service just caught fire, said Berkeley CEO Josh Coates. The start-up has sixty employees, and has raised only $2m in one round of VC financing. We don’t have a $1m marketing budget, Coates said.

But if EMC was mostly attracted by Berkeley’ back-end technology, the storage giant is not letting on. Berkeley’s technology is one of the most advanced in the industry, said an EMC spokesman. But he added: We were impressed with the entire online delivery model, not just one component of the business.

Berkeley and EMC both confirmed that the Mozy service is being used only for laptop and desktop backup, or for server backups at small businesses. But that does not mean that the Berkeley systems could not support backup services for servers at larger customers.

The front-end client software that Berekeley has developed and which automatically sends back-up data over the Internet to the company’s data centers will work just as well on bigger servers, Coates said. Actually it’s notoriously difficult to write front-end backup software for the high-end, for say Exchange for SQL servers. Our software already does that, he said.

Berkeley implicitly credits Coates with the design of its grid. The CEO has had practice building large scale storage at the now defunct storage service provider Scale8, and at a non-profit organization that was aiming to archive all human knowledge.

Earlier this year the grids in Berkeley’s two Utah data centers totaled 1PB in data capacity, but they now span 3PB of disk. Last week Berkeley told Computer Business Review that the system uses a Reed-Solomon encoding algorithm to protect data with much lower overheads than conventional RAID striping.

Yesterday Coates was tight-lipped about his grid, and said he was under orders not to talk about it. It’s not a NAS or a SAN, and it’s based on a completely different type of software architecture, he said. It does not use InfiniBand, and it is based on low-cost hardware. Is there any EMC disk in there? Oh no, not at all, he said with absolutely no hesitation.

Coates said that he could have attempted to sell a productized version of his grid. But that’s a much harder business, and it would have needed much more VC money than we wanted to raise, he said.