Cloud computing is taking off in large part because the delivery of IT services and infrastructure over the Web offers better pricing and more flexible pricing options. That, at least, is the theory. But what if cloud pricing models are so confusing that IT shops end up paying more than they expect?
These issues related to cloud pricing models were tackled at Interop Las Vegas Tuesday morning in a panel of vendors moderated by Allan Leinwand, a venture partner at Panorama Capital.
"The reason cloud is compelling is the pricing levels," Leinwand said. But the cloud industry has come up with several types of billing options, and they're not always easy to understand.
"You're talking about units that people don't normally think about," Leinwand said. "CPU hours: that's not something I go buy. I buy a blade server, and the hours are infinite, they're mine."
Even if an IT pro finds it easy to understand CPU hours, a CFO might not. "Try to explain to your CFO how many CPU hours you're going to use in the cloud, and see if they care," Leinwand said.
Pay-as-you-go models sound great at first, because they could eliminate long, fixed contracts that may force IT to pay for more services than they actually use. Panelist Richard Dym, chief marketing officer of vendor OpSource, for example, says his company is offering hourly billing with no long-term commitments.
System administrators "want to try stuff prior to having signed contracts, having a legal review," said Jesse Robbins, co-founder and CEO of start-up Opscode. "The long lead, enterprise sales cycle is something where I would rather have to eat glass than deal with it. It's the most painful thing imaginable."
But the pay-as-you-go model introduces the risk that a customer uses far more resources than it expects, leading to a high bill, an event experienced by many cell phone customers when they exceed their allotted minutes. Such a problem is especially likely if a company lacks visibility into its usage of cloud services.
Amazon's Web Services don't expose how much a customer owes in real time, said Thorsten von Eicken, CTO and founder of cloud management vendor RightScale. That's why RightScale has built automation tools that, among other things, show customers exactly how many Amazon minutes and hours they have used so they can gauge what to expect in their bill at the end of the month.
"It's really a question of visibility and control," von Eicken says.
The cloud won't totally eliminate the prospect of long, fixed contracts. Panelist Grace Kim, senior marketing manager of Cisco's WebEx, said WebEx's typical offer is a one-year contract. But Kim spoke about the need for a flexible model that lets end users do their jobs without worrying about the bill their company might get at the end of the month. "It's very important to choose a model where you're not penalizing end users for doing their jobs," she said.
One important consideration for enterprises considering cloud services is the service-level agreement, and whether it is easy to collect refunds if the cloud provider doesn't meet its obligations.
Dym was asked by an audience member if billing systems are being connected with service-level agreements to provide automatic refunds. He answered "I don't believe so."
Leinwand said he expects cloud billing models to evolve over time, just as long-distance phone services have gone from a pay-per-use model to a mixture of pricing structures that includes flat rates. But the transition could be rocky, he said.
"My assertion is cloud is new enough," he said. "Adding yet another billing model and consumption model [could] add more friction to the process for enterprise deployments."