University essay from Umeå universitet/Institutionen för datavetenskap

Author: Daniel Harr; [2019]

Keywords: ;

Abstract: Many modern soft‰ware systems are designed into a microservice-oriented architecture as they run into issues when a‹ttempting to scale. An issue with large and complex microservice-oriented systems is to know which scales of a system that are well-performing with regard to resource usage. Identifying effcient scales is interesting to minimize resource usage and cost while maximizing performance.‘ The optimal scales of a demo system is investigated using multi-objective Ant Colony and Particle Swarm optimization. Th‘e optimization methods are evaluated and compared with respect to properties of the resulting set of scales, and how much of the search space that is discovered for the solutions to be produced.‘ The experiments show that Ant Colony is more consistent in producing the entire correct set of scales. Particle Swarm however is cheaper with regard to the number of scales that need to be tested in order to produce a result. Since testing a scale becomes more expensive as the investigated system grows in size and complexity, an initial conclusion is that Particle Swarm would be more viable for a real-world scenario. ‘There are however some ideas of improvements that could a‚ffect the conclusions, and a larger and more complex system should be tested as well before any real conclusions can be made.

  AT THIS PAGE YOU CAN DOWNLOAD THE WHOLE ESSAY. (follow the link to the next page)