"I would like to contribute to making distributed data stores faster and more reliable"
Ying Liu studied on the European Master in Distributed Computing, one year in Barcelona and one year at KTH. After his studies he was enrolled as a doctoral student at the Erasmus Mundus Joint Doctorate programme and now he is halfway through with his newly defended licentiate thesis "Towards Elastic High Performance Distributed Storage Systems in the Cloud".
Where are you from and where did you study before coming to KTH?
– I am from China, I have finished my master degree from the European Master in Distributed Computing programme with one study track at UPC in Barcelona, Spain and one at KTH.
What is your topic and why did you choose it?
– My research focuses on achieving high performance in distributed storage systems. I choose this topic because I am particularly interested in distributed data stores and would like to contribute with my efforts to make them faster and more reliable.
Describe your topic in short.
– Nowadays, most Internet-based applications use back-end storage systems to store application and user data. Usually, the performance of these back-end storage systems significantly affect the performance of the applications. My topic is to improve the performance of back-end distributed storage systems in order to provide better user experience, low latency in particular, of Internet-based applications.
Tell us something about your results.
– Providing low latency storage solutions has been a very active research area with a plethora of open issues and challenges to be addressed. My work has contributed in providing low latency storage solutions by investigating solutions for the emerging novel system usage scenarios, i.e., global distribution, and by automatic scaling the system according to dynamic workload patterns.
What will the future bring for your research, how will you continue and what focus will you have on your PhD thesis?
– My research outcomes have the opportunities to be continued and improved in many ways. For the work to design low latency storage solutions in a global scale, I am particularly interested in data consistency algorithms that are able to provide the same consistency level while requiring less replica synchronization. Also, performance interference might happen when storage solutions are hosted in a virtualized environment (Cloud computing), which is usually the case. I would like to broaden my work in this direction by trying to identify performance interferences and investigate solutions to prevent or alleviate the effect of performance interference. Lastly, I would pursue my work in designing auto-scaling controllers for distributed storage systems focusing on improving their control accuracies.