One simple lesson learned from Amazon.com Inc.'s outage last week at its Elastic Compute Cloud (EC2) data center in Virginia is that technology fails. "It's not a matter of 'if,' but 'when,'" said Douglas Menefee, CIO at Schumacher Group, a Lafayette, La.-based provider of medical informatics. As such, CIOs -- like those at EC2 customers Evite, Quora, Reddit and Foursquare -- need to develop backup and recovery plans for when the cloud goes down.
"Just about every cloud solution I’ve adopted has had a major outage early in their development," Menefee said. "[The providers] learned from those and are focused on not repeating history. If Amazon continues to have issues … CIOs will not trust their solution."
Most midmarket companies are using cloud services for backup and recovery, not for primary storage, according to Richard Csaplar, senior research analyst for storage and virtualization at Aberdeen Group Inc., a research firm in Boston. Thus, the recent Amazon outage had less impact on companies unable to back up data than it would have on businesses running applications based on primary data in the cloud. Latency and bandwidth still need to be addressed before companies use the cloud for primary storage, Csaplar said.
"I would still match the uptime Amazon provides to that which small and midsized and even larger organizations can generate for themselves in their own IT departments," Csaplar said. "I know it is easy for me to say, but we need to remember that the concept of cloud and using it to replace or augment private computing resources is still very recent. It is not surprising to have a hiccup like this, and it serves as a reminder to IT to always have a plan B, even for plan Bs."
Still, several of Amazon's EC2 customers this month remained dark for days after the initial outage on April 21 -- an unconscionable duration, according to Ed Bell, interim CIO for the Massachusetts House and Senate in Boston.
"Twenty-four hours is a very long time," Bell said. "Amazon was promising 99.5 availability, and they might still make that, but [CIOs] need to get much more granular -- asking what does 99.5 mean, and for how long?" If a cloud provider doesn't provide the service promised, "there should be stiff penalties." On the other hand, the incident might justify higher costs for better levels of service, Bell said.
Amazon EC2 outage offers a cautionary tale for CIOs
While many observers point fingers at Amazon, the episode serves as a warning for CIOs of businesses that rely on the "seemingly infinite resources" available via cloud providers, according to Lauren Whitehouse, senior analyst at Enterprise Strategy Group Inc., an analyst firm in Milford, Mass.
It is not surprising to have a hiccup like this, and it serves as a reminder to IT to always have a Plan B, even for Plan Bs.
Richard Csaplar, senior research analyst, Aberdeen Group Inc.
"Sure, [the cloud] is cheap and supports scale, but what about reliability? Every company -- large and small -- needs to plan for the inevitability of some type of interruption of compute, storage or networking services," Whitehouse said. Of Amazon's woe begotten customers, she asked, "What was their contingency plan? If they were hosting their own businesses, wouldn’t they have had a Plan B in place?"
Companies need to lobby for better service-level agreements (SLAs) with their cloud providers, Whitehouse said. "Today, [customers] are really lacking provisions for availability of services and, importantly, access to data. How will a company be compensated when SLAs are not met? This is key because some can suffer significant financial losses and other adverse business impact." On top of that, businesses should put a backup and recovery plan in place, with out-of-region instances or copies to add another layer of protection for uptime, she said.
Regardless of inevitable outages, cloud computing will continue to be adopted because it provides clients with economies of scale, according to Bell. "Businesses will look at their operations and stick with their core competencies, outsourcing everything else," he predicted.
Menefee agreed that the benefits will continue to encourage CIOs to adopt cloud solutions. "The business pressure of more -- at cost reduction and in compressed timelines -- will not allow us to go back to the bloated data center model for high-growth companies," he said. "The Amazon outage is a great reminder that we all need to continuously evaluate business continuity… even with cloud solutions."
Let us know what you think about the story; email Laura Smith, Features Writer.