High Availability Framework for Mix-Cloud Secure Applications

Abstract

Having one of the services, such as web applications, databases or telephony systems, unavailable because of a single server failure is very annoying, yet very common issue, especially if the service is deployed on-premises. The simplest way to address it is to introduce redundancy to the system. But in this case the amount of physical machines needed will raise, while their efficiency will drop as most of the services do not use 100% of machine's capabilities. The better way to solve the service availability issue is to logically separate the service from the underlying hardware, balancing the load between instances and migrating them between the physical machines in case of failure. This way is much more effective, but it also contains a number of challenges, such as configuration difficulty and inter-service request routing.

The HA framework discussed in this thesis was designed to mitigate those issues. The key goal solved by the HA framework is raising the scalability and reliability of the service while keeping the configuration as simple as possible. The framework binds together a number of existing technologies, automatically installs and manages them with the single goal in mind: to provide an automated, easy-to-use, reliable, and scalable High Availability solution. In addition, the framework provides a distributed yet unified point of control over the whole installation, regardless of the physical location of components, including cloud and PaaS deployments. The framework is meant to be used by small-to-medium sized enterprises.