Introduction to ΔQ and Network Performance Science (extracts)
4 likes3,190 views
This document introduces δq, a groundbreaking concept in network performance science that redefines how packet-based networking is analyzed and measured. It emphasizes δq's role in providing a unified stochastic model for variability and loss, enabling robust predictions and performance manipulation in networks. The document also highlights the commercial significance of δq, asserting its value in optimizing network economics and aligning technical resources with financial management.
Introduction to ΔQ and Network Performance Science (extracts)
- 1. Introduction to ΔQ and Network Performance Science © Predictable Network Solutions 2014 INTRODUCTION + SUMMARY EXTRACTS
- 2. Dr Neil Davies Co-founder and Chief Scientist Ex: University of Bristol (23 years). Former technical head of joint university/research institute (SRF/PACT). The only network performance science company in the world. • High-fidelity network performance measurement and analysis. • Performance modelling and prediction for packet networks. • World’s first network contention management solution. PREDICTABLE NETWORK SOLUTIONS Peter Thompson CTO Ex: GoS Networks, U4EA, SGS-Thomson, INMOS & Universities of Bristol, Oxford, Cambridge and Warwick. Authority on technical and commercial issues of converged networking. Martin Geddes Associate Director of Business Development Ex: BT, Telco 2.0, Sprint, Oracle, Oxford University. Consultant on the emerging telecommunications technology and business models.
- 3. ΔQ AND NETWORK PERFORMANCE SCIENCE Overview of © Predictable Network Solutions 2014 3
- 4. Packet-based networking is a new discipline • It is fundamentally different from earlier forms of circuit-based telecommunications technology. • To reason about its performance requires new concepts – in addition to well-understood old ones. • Using the right conceptual tools transforms performance engineering from a skilled craft into a mathematical science. 4 ∆Q is the foundational concept for this network performance science.
- 5. What is ΔQ? • ΔQ is a fundamental breakthrough in the mathematics of stochastic systems. – A conceptual innovation like ‘zero’ in arithmetic, or ‘imaginary numbers’ in complex analysis. – A new branch of mathematics that underpins probability theory. – It has comparable utility to the development of computability in the 1930s or information theory in the 1940s. • ΔQ is a unified stochastic model of variability (typically of time delay) and loss. – Standard queuing theory, when applied to networks, fails to model reality, since it inadequately models loss. – The mathematics of ΔQ is based on improper random variables and their composition, in order to model loss. 5
- 6. Network performance Customer experience CONCRETE CONCRETE • A morphism is a model that can be viewed from multiple levels of abstraction. • This morphism formally relates network performance to customer experience • ΔQ is the only possible general network measure that is: – both a strong proxy to application outcomes (hence the customer experience) – and a network performance metric ΔQ is a morphism 6 ABSTRACT ΔQ ΔQ ΔQ ΔQ ΔQ ΔQ ΔQ Model that joins these
- 7. We use ΔQ to measure, model, and manipulate performance 1. ΔQ-based measurement is a universal quality of experience (QoE) proxy – ΔQ can also be composed & decomposed in several ways, depending on the level of abstraction used. 2. ΔQ can be used to predictively model network performance. – Robust prediction with strong philosophical foundations is the essence of the scientific method. 3. The ΔQ approach enables powerful new ways to manipulate packet network performance. – Technologies based on ΔQ enable networks to be driven to their theoretical mathematical limits. 7
- 8. The value of ΔQ and performance science • ΔQ helps us to reason about the performance of complete distributed computing systems. – It offers a new perspective that is focused on the computational outcome, not the packets per se. – It offers both prediction and assurance. This a (missing) critical precondition to successful network virtualisation. • Techniques using the ΔQ approach can transform network capability and cost – to enable sustainable broadband network economics. 8
- 9. WHY IS ∆Q IMPORTANT? Summary © Predictable Network Solutions 2014 9
- 10. Why is ∆Q important? • Fundamental technical breakthrough – Similar to Maxwell’s wave equation (continuous); Boltzman’s kinetic theory of gases (discrete) – Key features: power laws; error of measurement; system of 2 degrees of freedom – Conservation law (“=”, “>”) – like E – mc2 = 0 • Aligns technical with commercial – Captures dynamic vs static properties – QoE is about making bad experiences rare and ∆Q models failure modes – Allows optimisation to a cost/QoE goal © Predictable Network Solutions 2014 10
- 11. Why is ∆Q commercially valuable? • ∆Q is a rational unified resource model – Commercial resources: money, QoE – Technical resources: resource opportunity costs, ∆Q • It is a fundamental conceptual advance in our ability to financially model networks – Foundational idea like ‘double-entry book keeping’ in economics, or ‘just-in-time’ in manufacturing • Allows performance budgeting – Aligns technical and financial management – Enables prospective cost models, and not merely retrospective ones © Predictable Network Solutions 2014 11