Cloud outages can affect critical services
Most cloud application outages are caused by sporadic operations such as the deployment, upgrade or reconfiguration of applications. Insecure deployment pipelines and very high frequency of releases exacerbate the problem.
It is difficult for cloud users to understand the causes of outages, or to predict and control them, because they have limited visibility and control over the cloud environment.
In addition, an existing class of tools based on 'log analysis' currently fail to understand the behaviour of applications, and limiting their ability to detect or pinpoint errors.
Treating operations as processes
Our team has developed a method that uses 'process oriented dependability' (POD), which treats operations as processes and uses 'process context' to provide the basis for near real-time error detection, diagnosis and recovery.
We've started releasing POD tools to the developer community and continue our research on development of systems utilising POD as the core methodology.
In November 2014 our cloud migration and disaster recovery spin-out, Yuruware, which provides data protection technology for cloud-based companies such as Amazon Web Services (AWS), was acquired by Unitrends.