Tuesday, May 05, 2009

Moving Data around the Clouds

A startup hopes to make it easier to hop between cloud-computing services.

By Kate Greene

In recent years, companies led by Amazon, Microsoft, and Google have helped usher in the era of cloud computing, in which businesses and individuals lease access to computing on demand, paying only for as much processing power as they need. And as often happens when a new industry emerges, there's been a flurry of startup activity in the cloud-computing industry over the past year or so.

One startup, called Cloudkick, hopes to provide a simpler way to manage data stored across several different cloud-computing services. Cloudkick provides a unified, Web-based interface for monitoring data regardless of the cloud provider hosting it.

Another feature launched recently by Cloudkick, called Cloudshift, lets customers transfer data between different cloud-computing providers with just a few clicks. It makes it possible to shift an application from Amazon's servers to those of competitors, such as Rackspace, with surprising ease. This means that businesses using Cloudkick can avoid being locked into one provider--a feature that could help save money if a different provider suddenly offers a cheaper service.

"A lot of companies are afraid to move into the cloud because they don't want one company to have all their data," says Dan Di Spaltro, cofounder of the startup. These companies worry that it will take significant time and resources to move data between providers because, as yet, there is no cloud-computing standard. Moving from one provider to another requires technical tricks, including converting data between file types used by different providers. "We tackle that interoperability problem," says Di Spaltro.

Cloudkick presented its product last month at the Under the Radar conference in Mountain View, CA, where it won a best-in-show award and best in category, by audience vote. The startup, founded by the venture firm Y Combinator, came out of stealth mode in March, has 1,600 customers, and manages about 12,500 servers. The company's management service and its new migration service--currently only available to a small number of customers--are both free, although it intends to launch payment plans in the future.

After signing in to the Cloudkick website, a user can add different cloud-computing accounts by entering the necessary log-in information. A dashboard then shows each of these services and the status of the servers being used. A person can set up e-mail alerts to warn if traffic dips below a certain level or if it spikes, for example. The company also provides graphs to visualize the average load on a machine--an indicator of the overall health of the system.

Cloud control: This graph shows load averages for several machines. It can alert a user if the system is nearing failure or if an application is behaving poorly.
Credit: Cloudkick

Moving data from one vendor to another is as simple as dragging and dropping an on-screen icon. But behind the scenes, there's much more going on. When data is stored on Amazon's servers using its Simple Storage Service (S3), for instance, it is saved in a proprietary format that keeps it from simply being moved to another service. To solve this problem, Cloudkick has written software that automatically unbundles the data from the S3 file system. This involves removing a layer of encryption that surrounds the data, explains Alex Polvi, cofounder of Cloudkick. The data can then be uploaded to another provider--the whole process takes about five minutes. "It's like ripping a CD," says Polvi.

Cloudkick isn't the only company offering novel ways to manage cloud-computing services. Companies including CloudStatus and RightScale also provide a way to monitor cloud servers via the Web. Cloudkick's advantage, according to Polvi, is that it keeps things as simple as possible and now gives users the ability to switch servers so easily.

Di Spaltro adds that eventually Cloudkick would like to let customers transfer between cloud providers based on preset pricing options, or based on the geographic location in which the heaviest load occurs. For instance, if one provider offers a deal below a certain price, Cloudkick could automatically transfer data over to that service. Or if a customer's servers experience a heavy load in one part of the world during a certain time of day, he or she could automatically transfer the load to servers in that location at that time, reducing lag and improving performance.

"The value proposition to moving workloads around is great," says Frank Gillett, an analyst at Forrester Research. He says that the portability Cloudkick offers will appeal especially to startups, "who are shopping around for the lowest price or want to move quickly."

However, Gillett adds that there's still a lot of uncertainty about cloud computing and the types of services that will be the most useful. "The concepts are getting ahead of the market need," he says. "It's cool to see these startups springing up as experiments, but let's remember: we're at the very beginning."