Boštjan Merčun (2016) Improving reliability and performance of a ticket shop system. EngD thesis.
Abstract
At a time when Internet services are being used in most areas of life, it is very important that these services are always available. Users have become accustomed to highly available services to such an extent that they expect internet services to be available at any time, resulting in no tolerance of downtime. Each downtime can mean loss of customers and therefore of profit. In this thesis we will show an example of an architecture needed for selling tickets online. First we will look at the definition of high availability and explain why it is important and how we can achieve it. Then we will show the basic components of such architecture, define their drawbacks and analyze the reliability of individual components which we will later improve in order to prevent longer downtimes and consequently loss of customers. Highly available architecture will be further improved to handle high number of parallel users. When events are sold at a certain time, more users can visit the site than the system can handle. In order to avoid crashing the system, we will add a waiting room to the architecture which will ensure that web servers only get as many users as they are able to serve. At the end, we will evaluate the final architecture and evaluate how we tested and measured the improvements.
Actions (login required)
![[img]](http://eprints.fri.uni-lj.si/style/images/fileicons/application_pdf.png)
