One of the great things about doing your computing in the cloud is that you don’t have to worry about which machines your jobs are running on, or whether they have the right software on them, or even what city they’re in. Indeed, that’s the whole point. But one of the less ideal things about doing your computing in the cloud is that the network everyone uses to reach various public clouds—the Internet itself—is so unpredictable.
Cloud providers—who essentially rent out processing and storage resources so that companies can outsource IT infrastructure—can guarantee a certain level of service within their clouds. But there’s almost nothing they can do about traffic delays on the Internet, a problem that falls right back into the laps of users who were trying to increase efficiency in the first place.
Now, companies that rely on big cloud providers like Google, Amazon Web Services, or GoGrid will have a better way to see how the network connections linking them to each provider’s data centers are performing—and make judgments over time about which clouds are easiest to reach. That’s thanks to Apparent Networks, a Wellesley Hills, MA-based network performance software startup that’s introducing a service today called the Cloud Performance Center. It’s a free online tool that visually quantifies network performance for up to five “paths” between a user’s location and any specified cloud provider. (After the first five paths, Apparent charges $5 per path per month for the information.)
Information about network performance is useful because large delays or “latency” can disrupt business and weaken the argument for outsourcing computing jobs to off-premises resources. “Providers like Amazon or Google are building these beautiful data centers with top-notch people that have got these great services that are highly available, and that’s all good, but when I have to connect, I have to do it over the same old Internet,” says Jim Melvin, Apparent Networks’ president and chief marketing officer. “What we’ve done is provision the Internet, in North America to start with, with performance monitors in about a dozen key points like Boston, Miami, San Francisco, Vancouver, and Detroit. Our service will allow customers to see the performance for any of the top-tier cloud providers in those locations, trending back in time.”
So if you have an office in Boston, you can use the Cloud Performance Center to see which cloud provider has a history of the fastest network connections to the Hub. Melvin emphasizes that Apparent is “not trying to poke anyone in the eye” over the latency issue, but rather to raise awareness. Network connection slowdowns are most often a result of logjams within the network itself, he says.
“The vision and the ideal of cloud computing is that you don’t have to worry about where the compute cycles are being generated, but we’ve seen countless times that the reality of the Internet today is that you cannot count on connectivity,” he says. “You do need to care what level of performance you’re getting across the Internet.”
The fact that cloud computing providers can’t or won’t take responsibility for network performance issues once information leaves the data center “is one of the big stumbling blocks for the growth of cloud computing in general,” Melvin says. His company’s so-called “path-based” performance monitoring is one way businesses can get around that problem, he argues—indeed, he says some Apparent clients are already using the service to monitors tens of thousands of paths.
