What If My Storage Cloud Turns Stormy?
Services that store enterprise data in a "cloud" on the Internet raise questions that organizations are just beginning to ask, but for all their limitations, they may be no more risky than on-site storage platforms.
The technology behind cloud storage, as well as cloud computing, lets enterprises tap into IT resources without regard to where they're located. So cloud computing typically means relegating primary or backup data to an undefined repository outside the enterprise rather than a local data center or a dedicated remote site. By using cloud storage services, organizations can save on capital expenses and on complex setup and administration tasks, proponents say. Putting data in the cloud can also make it accessible from more locations.
Saving work and expense is a key reason why cloud services are expected to grow in the next few years. Last year, 4 percent of worldwide IT spending went to cloud services, and by 2012 that figure will be 9 percent, according to research company IDC. Because of its cost and space requirements, data storage is a prime candidate for a cloud solution, and IDC predicts storage will grow from 8 percent to 13 percent of cloud spending in that same period.
Vendors are stepping up to fill that demand. Amazon.com made an early splash in cloud computing and now includes a storage service, called S3, in its offerings. Nirvanix started offering a cloud archiving and backup service in 2007, and some of the biggest names in storage, including EMC's Mozy unit and Seagate, have also entered the game.
Anyone who's pondering how to deal with a cloud storage provider is already ahead of the game, according to analyst Henry Baltazar of The 451 Group.
"Right now, people aren't really (looking out) for this, because they're still weighing out whether they want to use cloud storage or not. ... It's still a brand-new market," Baltazar said.
But some companies are getting ready to dive in. Some form of cloud storage is likely to be a part of the data management plan now taking shape at Adventist Health in Roseville, California. It could be something as simple as hosting videos by hospital chaplains on YouTube, said Adventist CTO Greg McGovern. It makes no sense for a health care company to develop its own video player and hosting platform when it can take advantage of the most popular one on the market, he said.
FreshBooks, an online invoicing company in Toronto, is a beta tester of Rackspace's Cloud Files storage service and expects to start using it soon to store copies of large document files. That will save the company from having to manage the files and infrastructure, which is not FreshBooks' area of expertise, said CEO Mike McDerment.
"Being able to outsource that is a huge advantage," McDerment said. "Any business owner needs to figure out what business they're in."
But handing off any IT function involves some loss of control, and storage raises particular worries. For many companies, information is the core asset, and if employees and customers can't get to it, business grinds to a halt.
"Nobody's perfect. The best clouds in the world have downtime," acknowledged John Engates, CTO of Rackspace, a 10-year-old hosting company that has a cloud storage service in beta testing.
So what happens if things go sour with the service provider that's holding on to your data? How easy is it to get the information back or move it on to another provider? There are some dangers that enterprises should prepare for, but it turns out cloud storage may not be as risky as it seems, and possibly no more troublesome than an in-house system, according to users and industry analysts.
If it does come to the point of changing service providers, there is likely to be work involved, said IDC analyst Benjamin Woo.
"It's much more involved than just saying, 'I don't like this provider. I'm going somewhere else,'" Woo said. For one thing, you may have to change the backup software you use on your own premises, if your new cloud provider's system doesn't support it. Internal policies and procedures may also have to change, he said.
As for the process of getting back the data held in the cloud or moving it on to another provider, there are no standards used in common across the industry, analysts said.
There's no equivalent in cloud storage of a common transport mechanism like SMTP (Simple Mail Transfer Protocol), said Joe Kvidera, founder and CEO of Procedo, which provides data migration software and services. Because the industry is still in its infancy, vendors are choosing their own compression, encryption and transport mechanisms to differentiate themselves, Kvidera said.
But some vendors are taking steps to give subscribers more control and make migration easier. FreshBooks' McDerment believes it would be easy to move his archived files to another cloud. Rackspace provides a Cloud Files API (application programming interface) that could be used to write a new script, he said. Meanwhile, the API allows FreshBooks to manipulate and arrange its documents just by changing code, he said.
Nirvanix also provides a set of APIs, and it has worked out integration deals with vendors of backup and archiving software such as Atempo so customers can keep using the utilities they're familiar with, said Nirvanix President and CEO Jim Zierick. In addition, Nirvanix offers CloudNAS, a bridging software that can make the Nirvanix cloud look like any NAS (network-attached storage) drive. Using the Nirvanix API set, it mimics commonly used file systems such as CIFS (Common Internet File System) and NFS (Network File System), Zierick said.
Still, migrating a large amount of data from one service provider to another, or from a cloud to in-house storage, is a major undertaking. The basic Nirvanix service uses the public Internet to move files around, but the company has helped individual customers to rent high-bandwidth lines to handle large one-time shifts to the Nirvanix cloud, Zierick said. It has even helped customers load the data on a server at their facility and then physically transport that server to Nirvanix for offloading. The company is now talking to storage consulting firms about offering migration services.
Procedo's Kvidera said demand for migration services, which can cost between US$5,000 and $40,000 per terabyte, is growing rapidly.
But worries about availability and being able to back out of your current platform are actually as old as data storage itself.
Adventist Health CTO McGovern remembers working with mainframes 20 years ago, when critical data was backed up to tape. Two copies had to be made of everything, because some tapes turned out to be unreadable, and the nightly backup process would tie up the mainframe for as long as seven hours, he said.
"The Internet storage model is more reliable, more cost-effective and more assured than anything we've done in the past," McGovern said.
Using in-house resources to maintain data gives a feeling of safety and quick availability, but that may be just an illusion, he said. A tape stored in a leaky company warehouse may be less retrievable than data stored in a third-party cloud. Either way, the IT department should go through regular drills to make sure archived data is usable, McGovern said.
Internal storage, one of the most proprietary areas of IT, still has its own costs and pitfalls, said Nucleus Research analyst Rebecca Wettemann.
"If all my data is in a proprietary storage architecture within my own architecture ... is it less difficult to get it out and move it to another vendor? The answer is probably no," Wettemann said. The costs are likely to be higher, too, because a company may have large investments in specialized training as well as hardware and software, she said.
In addition, enterprises are likely to have more leverage with cloud storage providers than with companies that sold them storage infrastructure, Wettemann said. If there's a problem, a service provider will have more incentive to make the customer happy because it needs the next month's subscription fee, she said.
If push really comes to shove, a cloud storage provider might hold a customer's data hostage. At Rackspace, it's never come to that, Engates said. Nirvanix includes language in its contracts that lets it block a customer's access to data, but the company would do everything in its power to resolve a dispute before it came to that, Zierick said.
IDC's Woo thinks serious conflicts between a cloud storage provider and a customer are unlikely because the service is so simple.
"The most optimal relationship to have in a backup scenario ... is a non-existing one," Woo said.