Performance and Scaling in E-Commerce Systems

This course enables its participants to understand the problems of performance and scalability of software systems. It presents the relevant business models, technologies, and architectures. Additionally it shows performance models and metrics and how they are used in capacity planning models and forecasting methods.

Course content: Models for E-business, Customer Behavior Models, E-Business Functions, Performance Models, Service Time and Demand Queues, Performance Laws, Queueing Networks, Cost of Security (Payment Systems, Cryptography, Firewalls, etc.), Benchmarking.

News

Exam dates for winter semester 2015/2016 cab be found on our main page.

The exam correction review will be on Wednesday, May, 13th in E302 at 15h-16h. Please note that you will get a chance to look at your corrected exam and take notes. You may then hand in a written application (on paper or by email) of any correction items you would like to have reviewed. The strictly enforced deadline for this application is Sunday, May, 17th at 23:59.

Exam

The exam correction review will be on Wednesday, May, 13th in E302 at 15h-16h. Please note that you will get a chance to look at your corrected exam and take notes. You may then hand in a written application (on paper or by email) of any correction items you would like to have reviewed. The strictly enforced deadline for this application is Sunday, May, 17th at 23:59.

Slides

Performance Slides : (PDF)
Scheduling in Server Farms (Keynote SIPEW 2008): (PDF) | (PPT)
Stateful Stream Processing in the Cloud (Talk by Peter Pietzuch): (PDF)

Contents

  • Motivation
  • Software Performance Engineering
  • Service Level Agreements
  • Checklist of performance factors
  • Framework for quantitative analysis in e-commerce
    • e-business model
    • functional model
    • customer behavior model
    • IT resource model
  • Application patterns
  • Customer Behavior Model, CMBG, metrics, examples
  • Customer Visit Model, metric, examples
  • Architectures and interaction modes
  • Client/Server Interaction Diagrams
    • answering quantitative questions with the CSID
    • workload characterization
    • network performance (service time and utilization)
  • Queues
  • Performance laws
    • forced flow law
    • utilization law
    • service demand law
    • Little's law
  • Application of performance laws to capacity planning
  • Modeling with queues
    • open models
    • closed models
  • Single queue: infinite population/infinite queue
  • Single queue: infinite population/finite queue
  • Generalized system level models
    • Infinite population, infinite queue, variable rate
    • Single queue, fixed service rate, finite population
  • Queueing Networks
    • single class open networks
    • single class closed QN
    • mean value analysis
    • multiple class open QNs
    • multiple class closed QNs
  • Software contention
  • The cost of security (performance penalties)
    • cost of cryptography
    • firewalls
    • authentication
    • electronic payment methods
    • LDAP servers
  • Benchmarks
    • ECPerf, SPECjAppServer, TPC-W

Bibliography

Recommended Books

  • The Benchmark Handbook, Jim Gray
  • Scaling for e-business, Menasce y Almeida
  • Capacity planning for Web Performance (Metrics, Models and Methods), Menasce y Almeida
  • High Performance Client Server, Loosley Douglas

Articles

Course Information

TUCaN-Link 20-00-0075-vl
Lecture Thu. 08:55-11:30
in S2|02 C120
CP (SWS)

5 (3)

Language English
Exam

tbd

Office hours Mon., 16:30-18:00; please register with Maria Tiedemann

Organizers

Prof. Alejandro Buchmann

A A A | Print | Contact | Legal note | Search