Treinamento em Análise Quantitativa & Planejamento de Capacidade. Virgilio A. F. Almeida

Transcription

1 Treinamento em Análise Quantitativa & Planejamento de Capacidade Virgilio A. F. Almeida DATAPREV Rio de Janeiro 27 Novembro de 2009 Módulo #3a Departamento de Ciência da Computação Universidade Federal de Minas Gerais - UFMG 1 Virgilio Almeida, UFMG, 2009

2 Virgilio Almeida, UFMG,

3 What is Software Engineering? A discipline whose aim is the production of quality software, delivered on time, within budget, and satisfying users' needs. The specification, development, management, and evolution of software systems. Designing and developing high-quality software Successful software engineers will constantly learn and adapt new technologies Source:

4 Software Engineering Today: Software Engineering An engineering discipline that includes these processes and products: Software Engineering Management Software Requirements Analysis Software Configuration Management Source: SWEBOK Software Design b k / Software Construction Software Testing Software Engineering Infrastructure Software Engineering Process Software Evolution and Maintenance Software Quality Analysis

5

6

7 Performance Engineering Analyze the expected performance characteristics of a system during the different phases of its lifecycle Predict and otimize performance Collection of methods for the support of the development of Performance-oriented oriented systems

8 PE Larger Questions How can one plan, design, develop, deploy, and operate IT services that meet ever increasing demands for performance, availability, reliability, and security? Is a given IT system properly designed and sized for a given load condition? Virgilio Almeida, UFMG,

9 PE Activities Understand d the key factors that t affect a system s performance. Measure the system and understand its workload. Develop and validate a workload model that captures the key characteristics of the actual workload. Develop and validate an analytic model that accurately predicts the system s performance. Use the models to predict and optimize the system s performance. Virgilio Almeida, UFMG,

10 Modeling Process 10 Virgilio Almeida, UFMG, 2009

11 Performance Engineering Basic Methodology Understand the key factors that affect performance Understand the workload Workload model Analytical l model Use the models to predict and otimize performance

12

13 Typical PE Questions Can the insurance claim system meet its performance requirements of sub-second response time when a natural disaster occurs (e.g., a hurricane). Is the infrastructure of a government agency scalable and can it cope with the computing demands of the new required online security mechanisms? Is the reservation system for cruise lines able to respond to anticipated peak of customer inquiries after a TV ad campaign? Virgilio Almeida, UFMG,

14 Performance Engineering Methodology Model-Based Methodology Workload Model Performance Models Performance Objectives

15 Modeling Process

16 Motivating Example - Call Center An automotive parts distribution company decided to develop and implement new call center applications. Goals: Better relationships with customers Customer loyalty y Ensure quality of service Improve efficiency and service performance Identify and explore sales oportunities.

17 Goals: Call Center Foster better relationships with customers, creating customer loyalty and ensuring quality service. Improve efficiency and service performance. Identify and explore new sales opportunities. Main Functions: Order status inquiry Shipment tracking Problem resolution status inquiry Requirements: sub-second response time and 24x7 operation. 17 Virgilio Almeida, UFMG, 2009

18 Motivating Example: Call Center Architecture

19 Motivating Example - Call Center Customer inquirements: i Status of an order Location of a shipment Problems, when one wants to know what is happening with a given order Service respresentatives will have the functions Historical tracking of all customer contacts Single view of custumer informantion Record of resolution of past problems Help functions

20 Motivating Example - Call Center Is the system design able to meet the subsecond response time for all functions? Can the performance be maintened after integrating the sytem? Is the system capacity able to handle up to 1,000 calls on the busiest hour and preserve the subsecond response time? How do failures in the database server affect the 24x7 avaialability goal? What is the impact of start offering Web self-services to customers?

21 Motivating Example - Call Center Requirement analysis stage define the resources needed to support QoS goals Workload Model What workload do we want to characterize? All functions the IT application receives from the representatives during an oberserved period Packet size distribution, interpacket arrival time Load high level description Eg.: functions submitted by the representatives

22 At the Requirements Analysis Workload definition: Phase Call center s view: Arrival rate of phone calls IT system s view: Functions received from the representatives. DB server view: SQL requests from the application server. LAN view: packet size distribution and interpacket t arrival time. Virgilio Almeida, UFMG,

23 Motivating Example - Call Center Requirement design stage What should be the system throughput to meet the subsecond response time requirement? The analysts work the assumptions Number of service representatives: 200 Busiest hour: 80% of the representatives working Think time (period between two consecutive submissions): 30 sec.

24 Motivating Example - Call Center Model - abstraction

25 Motivating Example - Call Center N active representatives in the system X 0 system throughput N = 200 x 0.80 Response time law: R = N/X 0 Z < 1 sec X 0 >= N/(Z+1) = 5.16 functions/sec

26 Motivating Example - Call Center System development phase Before developing the database, we need to know what shoud be the performance of the database server Forced Flow law: X DB = V DB x X 0 X DB is the database server throughput V DB is the average number of visists per function to the database. Each function access the 2.2 times X DB >= 2.2 x 5.16 = transactions/sec DB

27 Motivating Example - Call Center Operation stage Check if the performance objectives are being met. R call = R Appl + R LAN + R DB R call < 1 (performance requeriment) Consider that the DB server has one CPU and two disks.

28 Motivating Example - Call Center DB server receives (peak hour): 57,800 queries/hour = X 0 = 16 queries/sec. Each query needs: needs 50 msec of CPU performs 4 I/Os on disk 1 performs 2 I/Os on disk 2 Each I/O takes an average of 8 msec

29 Motivating Example - Call Center D CPU = 0,05 sec D disk1 = V 1 xs disk1 = 4x = sec D disk2 = V 2 x S disk2 = 2 x 0.08 = sec U CPU = D CPU x X 0 = 0.05 x 16 = 80% U disk1 = D disk1 xx 0 = x 16 = 51.2% U disk2 = D disk2 x X 0 = x 16 = 25.6% R CPU = 0,25 sec R disk1 = sec R disk2 = sec CPU will be the bottleneck of the DB server

30 Motivating Example - Call Center Evolution stage The company is considering to develop Web applications Security: unauthorized modifications of the DB information. Interface: authentication ti ti services Internal architecture: access control mechanisms to the database information

31 Motivating Example - Call Center Solving the performance model The response time at the DB server: Local queries: 1.14 Web queries: 3.85.

32 Model-Based Methodology Understand the System Description of the system architecture, components and goals. Characterize the Workload Identify the basic components of the workload Measure the System and Obtain Workload Measure the System and Obtain Workload Parameters

33 Computer System Lifecycle Functional requirements: what the system has to do and on what type of platforms. Non-functional requirements: how well the system has to accomplish its functions. Service Level Agreements (SLA) are established. In many cases non functional requirements have been neglected or In many cases, non-functional requirements have been neglected or considered only at system test time!

34 Computer System Lifecycle How will the requirements be met? -System architecture - System broken down into components - Major data structures, files, and databases are designed. - Interfaces between components are specified

35 Computer System Lifecycle Components are implemented. - some are new - some are re-used - some are adapted Components are interconnected to form a system Components should be instrumented as they are built

36 Computer System Lifecycle Concurrent with system development, as components become available (unit testing) Integrated tests are carried out when the entire system is ready. Often, more time is spent in testing functional requirements than in testing non-functional requirements.

37 Computer System Lifecycle Configuration parameters have to be set in order to meet the SLAs. e.g., TCP parameters, database poolsize, maximum number of threads, etc.

38 Computer System Lifecycle Constant monitoring to check if the system is meeting demands: - workload (peak periods, unusual patterns) - external metrics (user-perceived) - internal metrics (help to detect bottlenecks and to fine tune the system) - availability (external and internal) It may be needed to dynamically adjust configuration parameters

39 Computer System Lifecycle Systems may need to evolve to cope with new laws and Regulations (e.g., HIPPA) Systems may need to evolve to provide new functions (e.g., sale of downloadable MP3 music in addition to CDs) How are the IT resources going to cope with evolution in terms of SLAs?

40 Reference Model for IT Business Model: - number of branches - number and location of ATMs - number of accounts of each type - business evolution plans (e.g., mergers) Social Model - privacy policy - accessibility ypolicy

41 Trabalho em Aula 41 Virgilio Almeida, UFMG, 2009

42 At the System Design Phase What should the system throughput be to meet sub-second response times? 200 customer service representatives and 80% are working during the peak hour. Average think time of 30 sec. 42 Virgilio Almeida, UFMG, 2009

43 At the System Development Phase What should be the capacity of the DB server so that the performance goals are met? Each submitted functions requires 2.2 SQL calls on average. From the Forced Flow Law: X DB = V X DB = tps 43 Virgilio Almeida, UFMG, 2009

44 At the Operation Phase Assume DB server is a problem. Response times exceed sub second goal. Measurements during peak hour: queries/hour Each query needs 50 msec of CPU, performs 4 I/Os on disk 1 and 2 I/Os on disk 2. Each I/O takes 8 msec on average. X 0 = / 3600 = 16 queries/sec Service demands: Dcpu = 0.05 sec; Ddisk1 = 4 x = sec; Ddisk2 = 2 x = sec. 44 Virgilio Almeida, UFMG, 2009

45 At the Operation Phase (cont d) Utilization computations (Service Demand Law): Ucpu = Dcpu x X0 = 0.05 x 16 = 80% Udisk1 = Ddisk1 x X0 = x 16 = 51.2% Udisk2 = Ddisk2 x X0 = x 16 = 25.6% Response Time (Open QN Model) R R R R ' CPU D = 1 U cpu D = 1 U cpu 0.05 = = 0.25sec ' disk1 disk1 = disk 1 ' disk 2 = D disk 2 1 U disk 2 ' ' ' 0 = Rcpu + Rdisk1 + Rdisk 2 = = 0.066sec = = 0.022sec sec 45 Virgilio Almeida, UFMG, 2009

46 At the Evolution Phase Develop Web based interface. Security requirements mandate that new applications be developed for Web access (authentication, auditing, DB access control mechanisms). Local Web Arrival 16 1 Rate (tps) Service demands (sec) CPU Disk Disk Virgilio Almeida, UFMG, 2009