Can you trust your data to Amazon, other storage cloud providers?

Amazon, Nirvanix and others promise easy data storage and infinite scalability for Web companies, but perils such as service outages and security breaches can’t be overlooked

The Amazon S3 and Nirvanix cloud storage services promise easy data storage for Web companies, but perils can’t be overlooked.

Michael Witz, founder of online file-sharing site FreeDrive, knows the horror of that proverbial middle-of-the-night call: "The site is down."

He lived the nightmare last fall when a fiber link between a Web server and 6TB of stored customer data went poof. With no access to customer data, Witz knew the Carlsbad, Calif., company needed a better storage strategy.

"Building your own storage system - it's not easy. It requires special knowledge. It's not in our core competency," Witz says. "We had a choice. We could build out our storage infrastructure, outsource it to a data center, make a capital investment in our hard disks, or we could outsource it to [a cloud service provider] like Nirvanix or Amazon."

Five questions to ask before trusting your data to a cloud provider

Cloud vs. utility: What’s the difference? Listen now.

Deciding on the latter option was "really easy," Witz says. Indeed, cloud storage - or, as he calls it, "an Internet hard drive for companies" - is emerging as an attractive storage option for an increasing number of companies that depend on delivering services over the Web. That's because with cloud storage, data resides on the Web, too, located across storage systems rather than at a designated corporate or hosting site. Cloud-storage providers balance server loads and move data among various data centers to ensure that information is stored close - and so delivered quickly - to where it is used. Cloud-storage users typically don't know where their data is stored at any given time.

Big clouds

The best-known cloud-storage service is Amazon's 2-year-old Simple Storage Service (S3). Cloud storage also is available from start-up Nirvanix, which launched in October 2007; and now through Mosso, a Rackspace company that unveiled its offering early this month. Amazon remains tight-lipped about its cloud infrastructure, but Nirvanix says it uses custom-developed software and file-system technologies running on Intel-based storage servers at six locations on the United States' East and West coasts, as well as in Asia and Europe. By year-end, the company expects to expand that number to more than 20. Customer data is replicated in two or three locations. Mosso initially is delivering its storage cloud from Rackspace's Dallas data center, with another site in the United Kingdom likely to be added soon, the company says.

Geoff Tudor, Nirvanix co-founder, compares cloud storage to electrical service. After all, he says, when you turn on a light switch, you don't know exactly from where each individual electron originates. The same applies to stored data in the cloud.

Storage clouds at a glance

Amazon Simple Storage Service (S3)

Available since March 2006

Monthly fee: 15 cents per gigabyte of storage

Additional fees: 10 cents to 18 cents per gigabyte for data transfers

Maximum file size: 5GB

Nirvanix Storage Delivery Network

Available since October 2007

Monthly fee: Starts at 18 cents per gigabyte of storage

Additional fees: 18 cents per gigabyte for uploads and downloads

Maximum file size: 256GB

Note: Mosso's CloudF3 is in private beta. Cloud storage services reportedly are in the works at EMC, Google and Microsoft.

FreeDrive's Witz has been using the Nirvanix cloud storage since November. He says he likes that Nirvanix can convert videos to flash format automatically and send data directly from the cloud to a FreeDrive customer. Without cloud storage, all data would have to be relayed through FreeDrive's own Web server. Witz would not specify how much cloud storage costs him, but he says FreeDrive uses the cloud to store data from 180,000 customer accounts and the service comprises a substantial part of his operating costs.

Nirvanix customers pay 18 cents per gigabyte of storage, per month, plus another 18 cents per gigabyte uploaded and downloaded; for S3, Amazon charges 15 cents per gigabyte of storage, per month, plus 10 cents to 18 cents per gigabyte for data transfers. A customer might start out with a few hundred or thousand users, but if its Web site takes off, it "can scale up to a petabyte [of storage] without ever changing the application," Nirvanix's Tudor says.

After he lived through numerous outages with hosted provider Media Temple, scalability is what drew Theron Parlin, CTO of personal-finance social-networking site Geezeo, to cloud storage. "We can store 50 times our current 251GB, and never have problems. And, the price will stay low," he says of the Hartford, Conn., company's reliance on Amazon S3 (as well as on Amazon's Elastic Compute Cloud).

Amazon and Nirvanix are the leading cloud-storage providers, but many of the technology world's biggest names are focused keenly on cloud storage and cloud computing in general. Google appears ready to launch an online storage service informally known as GDrive. EMC reportedly is preparing a massive storage cloud with technologies code-named Hulk and Maui. IBM, meanwhile, has several cloud-computing initiatives under the Blue Cloud umbrella. IBM, however, sees itself as the "arms provider" for cloud-storage services, rather than a service provider along the lines of Amazon, says Clod Barrera, system-storage chief architect at IBM.

Caution in the clouds

All this activity has turned cloud storage into something of a buzzword; if used too broadly the term could end up referring to any type of Internet-accessible storage, analysts caution. Enterprises should think of cloud computing as massively scalable IT capabilities delivered to external customers using Internet technologies, and cloud storage as that which is allocated to cloud computing-applications, says Stan Zaffos, a Gartner analyst.

In addition, enterprises have to remember that as compelling as cloud storage is, it isn't without potential problems. Amazon S3, for example, suffered an outage for several hours in February that resulted in numerous customer Web applications going offline. Amazon attributes the outage to elevated numbers of authentication requests, and has addressed the incident by adding "significant" amounts of capacity to the authentication service and improving the system that monitors the proportion of requests that are authenticated, says Adam Selipsky, a vice president for Amazon Web Services. Amazon hopes to avoid a recurrence, but notes that no data was lost as a result of the outage because Amazon stores multiple copies of every object in multiple locations.

Enterprises also must consider the possibility that data could be stolen or viewed by people who are not authorized to see it. "Any time you let the data out of your computer room you're asking for trouble from a security point of view," says Joel Snyder, senior partner with Opus One and a Network World product tester.

Potentially the biggest worry is that a company's data could be stored next to a competitor's, says Andrew Reichman, an analyst with Forrester Research. "The storage industry isn't fully there yet with really secure multitenancy offerings," he says, adding that data on which many enterprise applications depend probably shouldn't be allocated to cloud storage. "Cloud-storage providers recognize they need to do this, but it's going to take some time for the technology to catch up," he adds.

In the meantime, analysts say, cloud-storage users need to be proactive about security, encrypting sensitive data and securing data transit with such technologies as SSL.

Are you ready for cloud storage?

Do you trust your programmers? A simple mistake - dropping the "s" in an https transfer, for example - can put your files at risk, Snyder says. "There is this danger of not properly disciplining yourself to use these products in a secure fashion," he says.

Remember that the storage provider has less to lose than the customer. "Most mission-critical, high-risk applications probably should not go to the cloud," says Andrew Reichman, an analyst with Forrester Research.

Decide what type of latency you're willing to tolerate. When relying on the Internet, you may sometimes experience slowdowns, Reichman notes.

Do a cost-benefit analysis comparing an in-house solution with cloud storage. "It's not a given that cost is going to be lower, because [cloud-storage providers] have a gross margin they have to achieve to remain profitable," says Stan Zaffos, a Gartner analyst.

< Previous story: Law firm dumps tape for cheaper, greener CDP | Next story: Eight technologies for better data storage >


Copyright © 2008 IDG Communications, Inc.

The 10 most powerful companies in enterprise networking 2022