'Best effort' insufficient for cloud
SINGAPORE--Until there can be one point of accountability for cloud services, "best effort" portions of a provider's SLA (service level agreement) will continue to be a sticking point for wider enterprise cloud deployments, notes an industry analyst.
Speaking at a panel discussion here Thursday, Camille Mendler, vice president of Yankee Group's global service strategies, said the lack of control over every point of delivery--from network uptime to service uptime--will leave points of failure that cloud providers cannot guarantee. For example, a SaaS (software-as-a-service) provider may be able to guarantee service availability but cannot guarantee Internet connection uptime.
Mendler said enterprises should peruse vendors' SLA fineprint, noting that "as is" portions of these SLAs are enterprises' risks to bear. Such risks could outweigh the overall benefits of deploying cloud computing, she added.
"The cloud is trying to replace traditional IT but with very few guarantees," she said. "Regulation did not anticipate the cloud, and it is unclear how laws will apply to data residing in the cloud and moving data between clouds."
This gap in the market opens opportunity for "trusted intermediaries", namely operators, to provide an interface to businesses, handling various aspects from the orchestration and service management, to billing, reporting and security of a cloud deployment, she explained.
Ideally, such intermediaries should have experience providing cloud services and will need to tie up with other vendors to offer a complete service, she said, pointing to Asian operators Singapore Telecommunications (SingTel) and Japan's NTT as "potential" intermediaries.
SingTel recently outlined its cloud plans, which include a platform to offer hosted apps to enterprises that will be supported on the carrier's locally-based infrastructure.
Fabrizio Civitarese, Asia-Pacific vice president of sales and marketing at Reliance Globalcom, said the lack of cooperation between vendors means responsibility often finds itself in "no man's land"--where vendors push the blame of downtime to each other.
Speaking to ZDNet Asia on the sidelines of the panel discussion, Civitarese said it is "quite inevitable" that the intermediary role goes to an operator, because application service providers cannot be responsible for network uptime.
Even a large vendor such as IBM, would be unable to provide a "meaningful" guarantee on Internet connectivity without a tie-up with an operator, he said.
On Amazon's upcoming launch in the region, Civitarese said the cloud giant, too, would need to establish collaborations with service providers and players based in the region.
"An application provider with no network cannot promise much," he noted.