Cloud adoption is widespread but a business being 100% in the cloud is rare and outages aren’t helping.
One of the reasons for this is a fear of a loss of control, the threat that if your IaaS goes down there is nothing you can do about it apart from alert your provider.
Some cloud experts are expecting a ‘second wave’ of cloud adoption, where businesses integrate it more deeply into their operations. However, with outages occurring on reasonably frequently, the deep embedding of services could be hindered.
The outages aren’t limited to small cloud providers, it’s happening to the big companies, which is a worry.
Fujitsu is the latest company to admit to an outage, with its Sunnyvale data centre in California being hit by a power outage.
Fujitsu, said: "On August 22, 2015 (Pacific Daylight Time), the public power provider to Fujitsu’s Sunnyvale facility experienced a major transformer station failure, causing a widespread power outage.
"As a result, Fujitsu’s Sunnyvale datacenter and the services it delivers (including the Fujitsu TPS5 public cloud, the Fujitsu private hosted cloud as well as other infrastructure services) were impacted. Fujitsu immediately notified customers and provided support to minimize impact."
Google has suffered various brown outs this year, in March a 43 minute outage hit its Cloud Compute Engine and AWS’s EC2 and S3 were both taken down in US regions in August and another hiccup occurred in July.
Simon Michie, CTO, Redcentric, told CBR: "High profile, public cloud outages can often have an negative impact on users’ perceptions of the reliability of the cloud for essential data, applications and services.
"In truth however service delivery is still higher than a typical mid-market enterprise would have if they used their own on-premise resources or infrastructure to deliver the same service to their users.
"The cloud does offer more for less and understanding that balance is key. The impact of outages on users really depends on what they’re using the cloud for.
"In a recent survey that we undertook of 200 UK IT directors, 63% told us that their ultimate cloud destination was to use a mix of selective cloud services, indicating that hybrid models will dominate in the future. This means that users are literally pick and mixing public and private cloud to match their needs for uptime and resilience."
Often the outages are short lived and typically less than an hour, but this won’t be particularly comforting to the customers affected.
Sometimes the outage is beyond the cloud providers’ control, as with Google’s outage on the 13th of August in Belgium. The local power grid was hit by four lightning strikes which knocked out power.
Google, said: "GCE instances and persistent disks within a zone exist in a single Google data center and are therefore unavoidably vulnerable to data center-scale disasters."
To avoid this, customers could switch operations to other Google Compute Engine zones, however, that may not always be easy for customers restricted by data location regulations.
Michie, said: "The key for users to ensure that an outage’s impact is minimised is to identify the level of protection that they need for critical data, apps, infrastructures or platforms. SLAs will provide reassurance of the level of uptime but talk to your provider first and identify what happens in the event of an outage.
"Also ask for evidence of independent assurance of the datacentres hosting the cloud service when choosing a provider. Ensuring that the technology hosting your apps and data conform to the strictest performance levels available should help to prevent frequent outages.
"Finally, check what disaster recovery service options are available and how they work. Users can mitigate outages by using disaster recovery services (even cloud based DRaaS) to replicate and switch over simultaneously if needed."
An area that providers need to work on is their communications regarding outages can often be lacking in detail from some of the large cloud providers. Although status pages can provide some insight, the lack of detail provided and lack of post mortems could lead to trust being lost.
If a company potentially loses millions due to a cloud outage then a detailed explanation from the provider should be expected.