There is a growing demand for dynamically scalable, yet cost-effective computing and storage capacity solutions. In response to these expectations the cloud architecture was developed, which is able to keep IT costs low yet provide a flexible solution and make autonomic computing available. The designing of cloud-based architectures requires caution and the precise knowledge of requirements.
Both private and public cloud solutions already exist as open source or proprietary systems, but public and private cloud co-operation is a new area. The first solutions providing co-operation appeared in the last few years. This co-operation opened unprecedented opportunities to increase the scalability of IT systems, and provided novel, and in many case more effective as well as quicker solution than the traditional approaches.
Within the frames of my thesis I have created a system running on hybrid cloud architecture. I present the capabilities, advantages and disadvantages of the hybrid clouds via this system. To accomplish the task I used the OpenStack open source cloud solution and HP Cloud public services. The connection between the clouds is established by an open source tool, called Brooklyn. The usage of Brooklyn makes possible to create systems running on multiple clouds by developing in Java language and following the DevOps method.
I carried out a number of tests with a scalable web application implemented on the top of this hybrid cloud. These tests were focused on latency and throughput. I managed to double the throughput with using hybrid cloud, but this was accompanied by an increase in average delay, too.