Cloud Computing & Hosted PBX News – Dallas, TX
Cloud Computing & Hosted PBX News – Dallas, TX

Cloud Computing Is Like An Airplane

A cloud computing service is like an airliner – even though it can experience devastating crashes that affect many people, like Amazon’s crash last month, it’s still safer than driving your own car … or IT infrastructure.

That’s the rationale provided by Simon Crosby, CTO of Citrix, during a panel discussion on cloud computing during a keynote address at Interop 2011. Also on the panel were Randy Rowland, senior vice-president of product development at Terremark; Andy Schroepfer, vice-president of enterprise strategy at Rackspace; and moderator David Berlind, chief content officer at UBM TechWeb.

Berlind began the discussion by reviewing the recent cloud outages and attacks at Amazon – up to three days for some users – and the Sony PlayStation Network, which lasted several days and affected 100 million customers. Berlind also quoted data that found that every one second of latency costs financial trading firms $100 million over a year.

“Is this mission critical?” Berlind asked of the cloud infrastructure in light of these outages. “This is a confidence issue.”

“It’s like an airline crash,” responded Citrix’s Crosby, “but it’s still safer in an airplane than driving to work.”

Crosby then wondered aloud what the cloud or IT equivalent is of the Federal Aviation Administration, the airline industry’s regulator and watchdog. Without waiting for an answer, he continued with his analogy. “Broadly, you’re far better off in the cloud than doing things your own way,” he said.

Outages should be expected in the cloud from time to time, said Rackspace’s Schroepfer. To minimise them, operators should invest in backup, redundancy and resiliency.

“We’ve got to be willing to spend money,” Schroepfer said. “IT does go down.”

Application writers need to fully understand the infrastructure they’re developing programs for, said Terremark’s Rowland. Amazon’s Elastic Compute Cloud (EC2) service allows multiple instances of customer applications to run in so-called Availability Zones within a region; but analysts say it doesn’t provide the necessary tools to load-balance applications between regions, so customers have to use additional software on top of their Amazon instances.

“I don’t feel sorry for the application writer that doesn’t understand the infrastructure they’re writing to,” Rowland said. “They need to understand how it’s built, do their due diligence.”

The panel then suggested customers may want to run applications redundantly in two clouds instead of one. This may increase costs but will still be less expensive than building, owning, operating and maintaining an exclusive infrastructure, they concluded.

“The cost is never equal to internal” IT infrastructures, Crosby said. “You may experience three days of outage but you wouldn’t have existed otherwise.”

Berlind then asked the panel if cloud providers needed to be more transparent in their infrastructure operations and service-level guarantees, and whether those SLAs needed to be more comprehensive and standards improved. The panel concluded that it’s up to the customer and provider to negotiate a contract beforehand that is detailed and airtight.

“They’ve got to develop that relationship, which includes transparency,” Schroepfer said. “Customers have to trust the provider; they’ve got to believe they are fixing (any glitch).”

“The contract defines that relationship,” Rowland concurred.

Crosby then compared the SLA discussion to Apple’s App Store for the iPhone, in which applications are stored and run in the cloud and consumer satisfaction seems to be high. But then he said that environment is very different from cloud-based IT for enterprises.

“Again, what’s the FAA” for enforcing cloud SLAs? Crosby asked.

Source

Brian