Continuous Delivery provides numerous opportunities for the automated delivery of software in test and production environments in various stages of the software delivery process. It includes manual activities largely because these are:
-
slowly, unrepeatable and unreliable
-
inconsistent>
on the different environments of time
- / p>
- intransparent, since they are carried out by an expert.
Photo: Dynatrace
This is complemented by concrete approach DevOps, originated from the English terms and Development Operations. DevOps represents cultural, organizational changes within the company and means that the departments development and operation – which previously worked mostly separated – to cooperate closely in the preparation and the successful use of software. This avoids that software is developed and tested without taking into account the practical requirements of the operation and performance problems or instabilities are not discovered until too late in the practical application.
DevOps means not only that employees in development and operation are informed each other of any changes, but also that new versions are automatically tested and installed. The approach also includes that all steps are measurable and therefore verifiable.
Because of the shorter feedback loops between the development and operation, there is significant potential for savings. But the majority of the time and cost reduction is due to the high degree of automation. In addition, the quality and performance of the software is at an early stage measured concretely, so that errors or problems previously identified and fixed.
Through a work based on DevOps Continuous Delivery Model for example, Amazon now has 75 percent fewer failures and 90 percent less downtime minutes. The provider performs on average every 11.6 seconds with a new Deployment. Only every thousandth still has a problem that can be corrected immediately with a quick roll-back. This is of course an extreme example and is not to say that hundreds of deployments are to strive for a day – it shows, however, what can be achieved by automating both in deployment and in the quality control
However, there are at Cintuous Delivery. some challenges: Usually load and performance tests are carried out immediately before the delivery of the software, with continuous delivery cycles in between but always shorter. So how can continuously secure the performance?
And on what basis the company decides whether or not the state of the software is fit for release or not?
No comments:
Post a Comment