SOFTWARE PERFORMANCE AND SCALABILITY

The student will learn advanced techniques on mathematical and computational optimisation, the fundamental knowledge for the design of software and systems for handling and operating with big data.
The student will know the fundamental aspects to be considered in the design of software with performance and scalability a low energy impact requirements.

At the end of the course, the student will be able to:
- identify the main performance patterns in a system: bottlenecks, congestion collapse, hidden congestion collapse, side effects of competition
- program a professional benchmarking tool and perform the experiments to assess the performance indices
- interpret the outcomes of a benchmark correctly
- model a software architecture by means of a queueing system or a queueing network
- indentify the bottlenecks and bounds for the performance indices in systems that consist of more components modelled by means of queueing networks
- compute the average performance indices in complex systems modelled by means of queueing networks

Basics of statistics and probability
- discrete and continuous random variables
- Markov processes

Server side programming in any language

Relational databases

Introduction to Performance Evaluation
- Performance patterns
- Performance indices
- Methodologies for the performance evaluation

Introduction to software/hardware architectures
- classification of the parallelism in processors (hyper threading, multi core, multi processor) and impact on the software performance
- RAID systems and their impact on the performance of software
- networks and impact on the performance
- review of software architectures

Benchmarking
- Design of a benchmarking experiment
- Benchmarking in open and closed loop
- Confidence intervals

Single queueing systems
- Kendall’s notation
- Little’s law
- M/M/1, M/M/m, M/M/Infinity queues. Erlang’s Formulas
- M/G/1 queues

Queueing networks
- Operational analysis
- Jackson’s queueing networks
- Gordon Newell’s queueing networks: convolution algorithms and Mean Value Analysis
- BCMP Queueing networks
- Using Java Modelling Tool (JMT) for the performance evaluation of queueing networks

Introduction to profiling of software

Case study: performance of applications based on bitcoin/blockchain

Henry H. Liu: Software performance and scalability. A quantitative approach. Wiley. 2009
Mor Harchol-Balter : Performance Modeling and Design of Computer Systems: Queueing Theory in Action. Cambridge press 2021

The exam can be given in two ways:
1) A written exam and an oral interview. The written exam consists of open questions on the course topics and exercises. The oral interview will be essentially a discussion of the written exam.
2) One lab project and an oral examination. The oral examination consists in the discussion of the lab assignments and questions about the topics of the course.

written and oral

The lecturer has a duty to ensure that the rules regarding the authenticity and originality of exam tests and papers are respected. Therefore, if there is suspicion of irregular conduct, an additional assessment may be conducted, which could differ from the original exam description.

In case 1), the exam assessment will take into account the ability to carry out and discuss the case studies proposed in the written exam (70%) and the ability to present the fundamental concepts of the discipline (30%).
The written exam and the oral interview will be evaluated according to this grid up to 100% of the maximum marking:
A) Basic knowledge of the subjects and its application: up to 70% (response times of queuing systems, bottleneck identification, operational laws, load testing)
B) Intermediate knowledge of the subjects and its application: up to 80% (understanding and application of optimal dispatching policies, product-forms, algorithms)
C) Advanced knowledge of the subjects and its application: up to 100% (understanding of the proofs of the results and ability to re-elaborate them for different scenarios)
In case 2), the exam assessment will consider the quality of the project (30%), the quality of the report describing it (20%), and the ability to present the fundamental concepts of the discipline (50%).

Laboratory and frontal lessons

This subject deals with topics related to the macro-area "Climate change and energy" and contributes to the achievement of one or more goals of U. N. Agenda for Sustainable Development