Outages Altering Cloud Perception, Practice

5 comments


It seems that every time a high-profile Internet service goes down, pundits predict that it will undermine confidence in cloud computing. And yet through 2010 the cloud model has shown surprisingly resilience in the face of this “headline risk,” continuing to gain momentum and mindshare.

But that may be changing. High-profile service outages at Amazon and Microsoft have made some prospects wary of placing business applications on cloud computing services, and have altered the way many existing customers deploy their services, according to analysts from The 451 Group.

“The world is not the same after the April 21st outage at Amazon and the August outages at Microsoft and Amazon,” said William Fellows, co-founder and Principal Analyst at The 451 Group, at the Tier 1 Hosting & Cloud Transformation Summit last week.

Outages Have Ripple Effect

In April Amazon Web Services experienced an extended outage that caused downtime or performance problems for man y social media services that rely on the company’s cloud computing services. The outage, which lasted as long as four days for some customers using Amazon’s storage services, was later attributed to a network configuration error. In August the European cloud operations of both Microsoft and Amazon were knocked offline by a power outage in Dublin.

The downtime has reinforced the doubts of many who were already skeptical of the cloud, but also grabbed the attention of a larger group of prospects who could be deterred by further outages. A survey of IT customers by ChangeWave found that 7 percent of users were “significantly” less likely to use cloud services, while 20 percent said they were “somewhat less likely” to adopt the cloud model.

Only a small percentage of those who were “significantly less likely” to use the cloud were currently using the model, said Paul Carton, the Vice President of Research at ChangeWave, which was recently acquired by The 451 Group. Carton said this suggests that the strongest reactions to the recent outages were seen among IT users who are skeptical about the cloud model and see these events as reinforcing their conclusions.

Prospects Focus on Reliability

But Carton said the more important group was the 20 percent who said they were “somewhat less likely” to adopt cloud services, as that group likely includes many prospects who are on the fence about whether to use cloud services. If these type of outages became a recurring phenomenon, Carton said, it could have a negative impact on adoption among this group.

Perhaps the biggest impact of the outage has been seen in how existing users approach cloud architectures, according to Fellows. “End users now want to mandate that they have multi-cloud strategies,” said Fellows.

That means the old-school concept of maintaining redundant infrastructure in multiple sites and data centers is being repeated in the cloud – albeit using virtual machines that are far easier to back up and deploy than physical hardware. This will make it somewhat more expensive for companies to run a resilient cloud architecture, but could boost business for cloud providers and create opportunities for “cloud brokers” like RightScale that can help end users shift data between cloud platforms.

While  noting the potential that further outages could deter some cloud prospects, The 451 Group remains decidedly bullish about the growth of cloud computing, projecting that cloud spending will grow from $1.4 billion in 2010 to about $8 billion in 2014.

Add Your Comments

  • (will not be published)

5 Comments

  1. If trends stay the same, costs of cloud computing will continue to go down and the reliability will increase. So far it's too early to say cloud computing is in decline.

  2. Just as you have redundancy in place inside your four walls, organizations should do the same for cloud computing infrastructures. Those Amazon customers with that foresight suffered virtually no outages as they had spread availabilities across multiple Amazon zones. In effect, they planned for disruptions which are ultimately inevitable in both on-premise and cloud computing environments.

  3. I personally believe that it is unethical to call something a "cloud" if it only resides in a single datacenter. Until the “cloud” infrastructure - whether it be compute, storage or otherwise - is spread across multiple datacenters with the ability to seamlessly replicate and fail over, it should continue to be called virtualization because that’s all that it really is. Very few competitive organizations agree with me today, but the vast majority of our enterprise customers agree 100% because a single datacenter “cloud” is not what they are looking for. Am I being too tough? Jonathan Hoppe President & CTO, Netriplex

  4. Chris Patterson

    I personally believe that it is unethical to call something a “cloud” when it is just SAAS under a different name. Ellison had it right in many ways during has big cloud rant. Too bad he directly contradicted himself and sold out. The idea of utility computing presumes a parallelism with utilities (water, electricity). Computing is far more diverse an ecosystem for it to broken down so simply. Like the old space shuttles, with some many moving parts, failures are bound to happen. How many have we had this year?