Photo sharing site SmugMug shifts from RAID to Amazon's S3 service

January 6, 2007

4 Min Read
Network Computing logo

Online photo sharing site SmugMug expects to save big bucks in storage hardware costs by shifting its data onto Amazon's recently launched S3 hosted storage service.

Amazon took the wraps off S3, which stands for Simple Storage Service, earlier this year. The offering is essentially a Web services interface to Amazon's own back-end storage and users are charged 15 cents for each Gbyte stored. (See Amazon Takes Aim at Hosted Storage.)

SmugMug, which is competing with the likes of Yahoo's Flickr, moved to S3 in the spring when it realized that its own storage farm was growing out of control. The firm currently has around 200 Tbytes of RAID-based direct attached storage on Apple Xserve devices, although this figure could leap to 500 Tbytes by the end of 2007, according to CEO Don MacAskill.

"With our current growth rate, that would be about $80,000 of hardware each month," he says, adding that, initially, he was on the lookout for cheaper hardware.

But the CEO was unable to find anything as cheap as S3. "Next year we will easily save more than $1 million and it could be more than that -- it just depends on our growth rate," he explains.The exec says that data center savings aren't just coming from hardware. "It's in terms of buying the RAID boxes, the power, cooling, the data center space, and the manpower needed to maintain them," he says. (See Blades Still Too Hot and Power Problems Plague Users.)

The exec told Byte and Switch that SmugMug spends around $23,000 a month with Amazon, and he no longer has to worry about issues such as power and maintenance.

SmugMug started shifting its data over to S3 about eight months ago, and Amazon now stores around 60 percent of SmugMug's 110 million photos. This accounts for about 100 Tbytes of storage.

It will be another three to six months before the rest of the photos are moved over to S3, although MacAskill says he is in no rush to shift the data: "When we moved to S3 we didn't want to copy everything over as quickly as we could, because then we would be left with idle hard disks that we have paid for."

Some of the Apple Xserve devices will eventually be deployed elsewhere within SmugMug for running database applications, and others could be resold. "We're excited about that, because it may mean that we will be getting money back for costs that we thought we had sunk," says MacAskill.SmugMug built its own software to replicate the data over to S3, partly to ensure that no data was lost in the transfer. The software, according to MacAskill, will not send each photo's metadata to Amazon until it is certain that the photo itself has been transferred.

The CEO admits there is an element of risk when you rely on hosted storage, particularly when it comes to availability. "It's something that we obviously think about all the time," he says, although he feels that Amazon is in an unusual position. "Amazon eats their own dog food -- when S3 has problems, Amazon has problems."

That said, the project has not been without hiccups, especially during a week-long beta test back in the spring. "Early on, we lost connectivity to them for 15 minutes, but it turns out that the whole of Amazon lost connectivity," explains MacAskill. "The whole [Amazon] company was on high alert, so we were back online really fast," says the CEO, stressing that there have been no similar outages since.

The exec, however, told Byte and Switch that there are some applications that he would not use S3 for, such as SQL server and Oracle databases. Because database I/O read/writes need to be extremely fast, these applications typically need to be close to servers. "[With S3] we're traversing the public Internet, which, as you know, sometimes has slowdowns and problems," he says.

Mountain View, Calif,-based SmugMug put its hardware in a nearby San Jose collocation facility which links to various ISPs' points of presence via OC-48 lines and higher, according to the startup. Amazon's data centers are located in Seattle and Ashburn, Va.The CEO, nonetheless, remains committed to hosted services. "We're looking at some of [Amazon's] other Web services," he says, such as the Elastic Compute Cloud (EC2), which offers hosted compute power. "We're already playing with it right now," he explains, adding that SmugMug may use this for image processing.

James Rogers, Senior Editor, Byte and Switch

  • Apple Inc. (Nasdaq: AAPL)

  • Yahoo Inc.

Stay informed! Sign up to get expert advice and insight delivered direct to your inbox

You May Also Like

More Insights