Scope quantification in Client/Server Environments Daniel A. Menascé George Mason University Fairfax, VA 22030 USA menasce@cs.gmu.edu
Slide 2Outline Part I: Client/Server Systems Part II: Introduction to Capacity Planning Part III: A Capacity Planning Methodology for C/S Environments Part IV: Performance Prediction Models for C/S Environments
Slide 3Outline (proceeded with) Part V: Advanced Predictive Models of C/S Systems Part VI: Case Study Bibliography
Slide 4Part I: Client/Server (C/S) Systems
Slide 5Definitions and Basic Concepts Client Server Work division amongst customer and server Client/Server correspondence
Slide 6. . . switch R LAN portion 1 FDDI ring switch LAN section 2 R . . . Definitions and fundamental ideas DB server DB server
Slide 7Definitions and essential ideas: Client Workstation with representation and handling abilities. Graphical User Interface (GUI) executed at the customer. Fractional preparing executed at the customer.
Slide 8Definitions and essential ideas: Server Machine with substantially bigger preparing and I/O limit than the customer. Serves the different solicitations from the customers. Executes a critical part of the handling and I/O of the solicitations created at the customer.
Slide 9Work division amongst customer and server Client Server GUI Processing Pre & Post Process. I/O COMM. COMM. DB correspondences arrange
Slide 10Interaction amongst customer and server Remote Procedure Call (RPC) customer DB server pre-proces-sing execute_SQL(par1,par2,...) server preparing result_SQL(...) post-proces-sing
Slide 11Part II: Introduction to Capacity Planning
Slide 12Migration to C/S case: "cutting back" a claim handling application DB server associated with a few PCs through an Ethernet LAN GUI application executing at the PCs LAN associated with the undertaking centralized server through a T1 line DB server is refreshed each night.
Slide 13Migration to C/S frameworks centralized computer based framework T1 line centralized server
Slide 14Migration to C/S DB server based framework DB server LAN portal T1 line centralized server
Slide 15Migration to C/S: some vital inquiries what number customers can be upheld by the DB server while keeping up a reaction time beneath 2.5 sec? To what extent does it take to refresh the DB consistently?
Slide 16Migration to C/S illustration: estimations with a model During 30 minutes (1,800 sec): 25% CPU usage 30% circle use 800 exchanges were executed Each exchange utilized: 1,800 * 0.25/800 = 0.56 sec of CPU and 1,800 * 0.30/800 = 0.68 sec of plate.
Slide 17Good News and Bad News Good News: we know the CPU and I/O benefit time of every exchange. Awful News: exchanges at the DB server go after CPU and I/O lines will shape at every gadget. We don't know to what extent every exchange sits tight in the line for the CPU and for the plate.
Slide 18cpu plate DB Server Model arriving exchanges withdrawing exchanges DB server
Slide 19CPU or I/O Times benefit request = 0.56 seg ? line holding up time
Slide 20Capacity Planning Definition Capacity Planning is the way toward foreseeing when the administration levels will be abused as an element of the workload advancement , and in addition the assurance of the most financially savvy method for postponing framework immersion.
Slide 21C/S Migration Example: wanted outcomes reaction time (sec) benefit level no. of customer workstations
Slide 22Part III: A Capacity Planning Methodology for Client/Server Environments
Slide 23Capacity Planning Methodology for Client Server Environments Understanding the Environment Workload Characterization Developing a Cost Model Workload Model Cost Model Performance Model Validation and Calibration Workload Forecasting Valid Model Performance Prediction Cost Prediction Cost/Performance Analysis Configuration Plan Personnel Plan Investment Plan
Slide 24Capacity Planning Methodology for Client Server Environments Understanding the Environment Workload Characterization Developing a Cost Model Workload Model Cost Model Performance Model Validation and Calibration Workload Forecasting Valid Model Performance Prediction Cost Prediction Cost/Performance Analysis Configuration Plan Personnel Plan Investment Plan
Slide 25Understanding the Environment Hardware and System Software Network Connectivity Map Network Protocols Server Configurations Types of Applications Service Level Agreements Support and Management Structure Procurement Procedures
Slide 26Example of Understanding the Environment 5,000 PCs (386 e 486) running DOS and Windows 3.1 and 800 UNIX workstations. IBM MVS centralized server. 80 LANs in 20 structures associated by a FDDI 100 Mbps spine. 50 Cisco switches. Arrange advancements: FDDI, Ethernet, T1 connections and Internet.
Slide 27Example of Understanding the Environment (proceeded with) Protocols being directed: TCP/IP and Novell IPX. Servers: 80% are 486 and Pentiums and 20% are RISC workstations running UNIX. Applications: office robotization (email, spreadsheets, wordprocessing), access to DBs (SQL servers) and asset sharing. Future applications: remotely coordinating, EDI, picture preparing.
Slide 28Capacity Planning Methodology for Client Server Environments Understanding the Environment Workload Characterization Developing a Cost Model Workload Model Cost Model Performance Model Validation and Calibration Workload Forecasting Valid Model Performance Prediction Cost Prediction Cost/Performance Analysis Configuration Plan Personnel Plan Investment Plan
Slide 29Workload Characterization Process of parceling the worldwide workload into subsets called workload parts . Cases of workload parts: DB exchanges, solicitations to a document server or, employments with comparable qualities. Workload segments are made out of fundamental segments .
Slide 30Workload Characterization: workload parts and essential segments
Slide 31Workload Characterization Basic Component Parameters Workload Intensity Parameters number of messages sent/hour number of question exchanges/sec Service Demand Parameters normal message length normal I/O time per inquiry exchange.
Slide 32Workload Characterization Methodology Identification of Workload Components Identification of Basic Components. Parameter Selection. Information Collection: benchmarks and ROTS (Rules of Thumb) might be utilized. Workload apportioning: averaging and grouping.
Slide 33Workload Characterization Data Collection Alternatives
Slide 34Benchmarks National Software Testing Laboratories (NSTL): servers and applications. Exchange Processing Council (TPC) System Performance Evaluation Cooperative (SPEC) AIM Benchmark suites
Slide 35Capacity Planning Methodology for Client Server Environments Understanding the Environment Workload Characterization Developing a Cost Model Workload Model Cost Model Performance Model Validation and Calibration Workload Forecasting Valid Model Performance Prediction Cost Prediction Cost/Performance Analysis Configuration Plan Personnel Plan Investment Plan
Slide 36Workload Model Validation
Slide 37Capacity Planning Methodology for Client Server Environments Understanding the Environment Workload Characterization Developing a Cost Model Workload Model Cost Model Performance Model Validation and Calibration Workload Forecasting Valid Model Performance Prediction Cost Prediction Cost/Performance Analysis Configuration Plan Personnel Plan Investment Plan
Slide 38Workload Forecasting Process of foreseeing the workload force. tps
Slide 39Workload Forecasting Business Units Number of business components that decide the workload development number of solicitations number of records number of representatives number of cases number of beds
Slide 40Workload Forecasting Methodology Application Selection Identification of Forecasting Business Units (FBUs) Statistics assembling on FBUs FBU estimating (utilize straight relapse, moving midpoints, exponential smoothing) and business vital arrangements.
Slide 41Linear Regression Example
Slide 42Capacity Planning Methodology for Client Server Environments Understanding the Environment Workload Characterization Developing a Cost Model Workload Model Cost Model Performance Model Validation and Calibration Workload Forecasting Valid Model Performance Prediction Cost Prediction Cost/Performance Analysis Configuration Plan Personnel Plan Investment Plan
Slide 43Performance Prediction Predictive models: investigative or recreation based. Explanatory models depend on Queuing Networks (QNs) effective take into account the quick examination of countless perfect for scope organization
Slide 44Performance Prediction considers that effect execution Client stations Servers Communication media Protocols Interconnection gadgets (extensions, switches and portals)
Slide 45Performance Prediction Model Accuracy
Slide 46. . . switch R LAN Segment 1 FDDI ring switch LAN portion 2 R . . . Execution Prediction An Example
Slide 47Performance Prediction QN for Example
Slide 48Performance Prediction Response Times for the Example Response Time (sec) Number of customers
Slide 49Capacity Planning Methodology for Client Server Environments Understanding the Environment Workload Characterization Developing a Cost Model Workload Model Cost Model Performance Model Validation and Calibration Workload Forecasting Valid Model Performance Prediction Cost Prediction Cost/Performance Analysis Configuration Plan Personnel Plan Investment Plan
Slide 50Performance Model Validation
Slide 51Capacity Planning Methodology for Client Server Environments Understanding the Environment Workload Characterization Developing a Cost Model Workload Model Cost Model Performance Model Validation and Calibration Workload Fo
SPONSORS
SPONSORS
SPONSORS
Rivalry Policy: Definition and Scope 
Focal Virginia Metropolitan Planning Organization 
Client Market Analyses for Focus and Validation 
Client Interface Programming in C: Direct Manipulation 
Taking a gander at Nearby Dailies' Decision 2004 Online News Scope Tania H. Cantrell Graduate Understudy School of News-
Client Guide 
How Training Supports an Organization s Strategy 
Man is not made for society, but rather society is made for man. No organization can be great which does not have a ten