Network Performance Measurement and Analysis

Similar documents
EVALUATION OF LOAD BALANCING ALGORITHMS AND INTERNET TRAFFIC MODELING FOR PERFORMANCE ANALYSIS. Arthur L. Blais

Internet Traffic Variability (Long Range Dependency Effects) Dheeraj Reddy CS8803 Fall 2003

Observingtheeffectof TCP congestion controlon networktraffic

Examining Self-Similarity Network Traffic intervals

CS551 End-to-End Internet Packet Dynamics [Paxson99b]

Overview of Network Measurement Tools

Connection-level Analysis and Modeling of Network Traffic

A STUDY OF WORKLOAD CHARACTERIZATION IN WEB BENCHMARKING TOOLS FOR WEB SERVER CLUSTERS

ON THE FRACTAL CHARACTERISTICS OF NETWORK TRAFFIC AND ITS UTILIZATION IN COVERT COMMUNICATIONS

Question: 3 When using Application Intelligence, Server Time may be defined as.

MEASURING WORKLOAD PERFORMANCE IS THE INFRASTRUCTURE A PROBLEM?

Network traffic: Scaling

Using IPM to Measure Network Performance

Web Server Software Architectures

IP SLAs Overview. Finding Feature Information. Information About IP SLAs. IP SLAs Technology Overview

Performance Measurement of Wireless LAN Using Open Source

Web Analytics Understand your web visitors without web logs or page tags and keep all your data inside your firewall.

Accelerated Simulation Method for Power-law Traffic and Non- FIFO Scheduling

Measuring IP Performance. Geoff Huston Telstra

Gaining Operational Efficiencies with the Enterasys S-Series

Using Fuzzy Logic Control to Provide Intelligent Traffic Management Service for High-Speed Networks ABSTRACT:

Optimizing TCP Forwarding

Network Management and Monitoring Software

Connection-level Analysis and Modeling of Network Traffic

pathchar a tool to infer characteristics of Internet paths

Communications and Networking

Comparative Analysis of Congestion Control Algorithms Using ns-2

Web Load Stress Testing

WEB APPLICATION PERFORMANCE PREDICTION

Network Performance Monitoring for Applications Using EXPAND.

Characteristics of Network Traffic Flow Anomalies

Towards Traffic Benchmarks for Empirical Networking Research: The Role of Connection Structure in Traffic Workload Modeling

Internet Management and Measurements Measurements

Cisco Bandwidth Quality Manager 3.1

Performance Evaluation of New Methods of Automatic Redirection for Load Balancing of Apache Servers Distributed in the Internet

Data Center Network Traffic Engineering and Statistical Design

Local-Area Network -LAN

A Statistically Customisable Web Benchmarking Tool

IP Network Monitoring and Measurements: Techniques and Experiences

Transparent Optimization of Grid Server Selection with Real-Time Passive Network Measurements. Marcia Zangrilli and Bruce Lowekamp

OpenFlow Based Load Balancing

co Characterizing and Tracing Packet Floods Using Cisco R

Hanyang University Grid Network Monitoring

First Midterm for ECE374 02/25/15 Solution!!

Evaluation and Characterization of Available Bandwidth Probing Techniques

Low-rate TCP-targeted Denial of Service Attack Defense

Traffic model and performance evaluation of Web servers

measurement metrics measurement techniques throughput measurement packet processing overhead packet size in real traffic

Lecture 8 Performance Measurements and Metrics. Performance Metrics. Outline. Performance Metrics. Performance Metrics Performance Measurements

Whitepaper Performance Testing and Monitoring of Mobile Applications

Delay, loss, layered architectures. packets queue in router buffers. packets queueing (delay)

Abstract. 1 Introduction

Network Traffic Monitoring and Analysis

CS640: Introduction to Computer Networks. Applications FTP: The File Transfer Protocol

Performance Analysis of AQM Schemes in Wired and Wireless Networks based on TCP flow

Internet Traffic Measurement

Intel DPDK Boosts Server Appliance Performance White Paper

A Simulation Study of Effect of MPLS on Latency over a Wide Area Network (WAN)

Introduction. The Inherent Unpredictability of IP Networks # $# #

Looking at the Server Side of Peer-to-Peer Systems

On evaluating the differences of TCP and ICMP in network measurement

Modeling users' dynamic behavior in web application environments

Quality of Service (QoS)) in IP networks

Testing & Assuring Mobile End User Experience Before Production. Neotys

architecture: what the pieces are and how they fit together names and addresses: what's your name and number?

APPLICATION-AWARE ROUTING IN SOFTWARE-DEFINED NETWORKS

DATA COMMUNICATOIN NETWORKING

An Adaptive Load Balancing to Provide Quality of Service

A Measurement Study of Available Bandwidth Estimation Tools

Effects of Interrupt Coalescence on Network Measurements

Probabilistic properties and statistical analysis of network traffic models: research project

LOAD BALANCING AS A STRATEGY LEARNING TASK

SiteCelerate white paper

APPLICATION LEVEL PERFORMANCE METRICS

Web Traffic Capture Butler Street, Suite 200 Pittsburgh, PA (412)

How To Monitor And Test An Ethernet Network On A Computer Or Network Card

Wave Relay System and General Project Details

Defending Against Traffic Analysis Attacks with Link Padding for Bursty Traffics

NETWORK BURST MONITORING AND DETECTION BASED ON FRACTAL DIMENSION WITH ADAPTIVE TIME-SLOT MONITORING MECHANISM

Characterizing and Modeling User Mobility in a Cellular Data Network

Hyper Node Torus: A New Interconnection Network for High Speed Packet Processors

IP Telephony Management

Advanced Computer Networks IN Dec 2015

A NEW TRAFFIC MODEL FOR CURRENT USER WEB BROWSING BEHAVIOR

Visualizations and Correlations in Troubleshooting

Choosing Application Performance Management (APM) Tools

allow all such packets? While outgoing communications request information from a

Lecture 12: Network Management Architecture

TE in action. Some problems that TE tries to solve. Concept of Traffic Engineering (TE)

An overview on Internet Measurement Methodologies, Techniques and Tools

Load balancing as a strategy learning task

Network-based Modeling of Assets and Malicious Actors

Master s Thesis. Design, Implementation and Evaluation of

An Empirical Model of HTTP Network Traffic

Lab VI Capturing and monitoring the network traffic

Cisco Configuring Commonly Used IP ACLs

G.Vijaya kumar et al, Int. J. Comp. Tech. Appl., Vol 2 (5),

How Router Technology Shapes Inter-Cloud Computing Service Architecture for The Future Internet

How To Model A System

Transcription:

Network Performance Measurement and Analysis Outline Measurement Tools and Techniques Workload generation Analysis Basic statistics Queuing models Simulation CS 640 1

Measurement and Analysis Overview Size, complexity and diversity of the Internet makes it very difficult to understand cause-effect relationships Measurement is necessary for understanding current system behavior and how new systems will behave How, when, where, what do we measure? Measurement is meaningless without careful analysis Analysis of data gathered from networks is quite different from work done in other disciplines Measurement/analysis enables models to be built which can be used to effectively develop and evaluate new techniques Statistical models Queuing models Simulation models CS 640 2

Determining What to Measure Before any measurements can take place one must determine what to measure There are many commonly used network performance characteristics Latency Throughput Response time Arrival rate Utilization Bandwidth Loss Routing Reliability CS 640 3

Measurement Introduction Internet measurement is done to either analyze/characterize network phenomena or to test new tools, protocols, systems, etc. Measuring Internet performance is easier said than done What does performance mean? Workload (what and where you re measuring) selection is critical Reproducibility is often essential Many tools have been developed to measure/monitor general characteristics of network performance traceroute and ping are two of the most popular These are examples of active measurement tools Passive tools are the other major category Representative and reproducible workload generation will be a focus CS 640 4

Active Measurement Tools Send probe packet(s) into the network and measure a response Ping: RTT and loss Zing: one way Poisson probes Traceroute: path and RTT Nettimer (Lai): latest bottleneck bandwidth using packet pair method Size/BW T1 T0 Tn+1 Tn Pathchar: per-hop bandwidth, latency, loss measurement Pchar, clink: open-source reimplementation of pathchar Problem: measurement timescales vary widely Tn+1 - Tn = max(s/bw, T1 T0) CS 640 5

Passive Measurement Tools Passive tools: Capture data as it passes by Logging at application level Packet capture applications (tcpdump) uses packet capture filter (bpf,libpcap) Requires access to the wire Can have many problems (adds, deletes, reordering) Flow-based measurement tools SNMP tools Routing looking glass sites Problems LOTS of data! Privacy issues Getting packet scoped in backbone of the network CS 640 6

Workload Generation Local and/or wide area experiments often require representative and reproducible workloads How do we select a workload? Currently HTTP makes up the majority of Internet traffic Trace-based workloads Capture traces and replay them Black-box method Synthetic workloads Abstraction of actual operation May not capture all aspects of workload Analytic workloads Attempt to model workload precisely Very difficult CS 640 7

SURGE Web Workload Generator Scalable URl Generator Analytic workload generator Based on 12 empirically derived distributions of Web browsing behaviror Explicit, parameterized models Captures heavy-tailed (highly variable) properties of Web workloads Widely used SURGE components: Statistical distribution generator Hyper Text Transfer Protocol (HTTP) request generator CS 640 8

Workload characteristics captured in SURGE BF EF1 EF2 Off time SF Off time BF EF1 Characteristic Component Model System Impact File Size Base file - body Lognormal File System * Base file - tail Pareto * Embedded file Lognormal * Single file1 Lognormal * Single file 2 Lognormal * Request Size Body Lognormal Network * Tail Pareto * Document Popularity Zipf Caches, buffers Temporal Locality Lognormal Caches, buffers OFF Times Pareto * Embedded References Pareto ON Times * Session Lengths Inverse Gaussian Connection times CS 640 9

SURGE Architecture SURGE Client System ON/OFF Thread ON/OFF Thread ON/OFF Thread SURGE Client System LAN Web Server System SURGE Client System CS 640 10

SURGE and SPECWeb96 exercise servers very differently Percent CPU Utilization 40 35 30 25 20 15 10 5 0-5 Surge SPECWeb96 0 200 400 600 Packets per Second SPECWeb96 SURGE CS 640 11

Analyzing Measured Data Analyzing measured data in networks is typically done using statistical methods Selecting appropriate analysis method(s) is critical Averaging Dispersion (variability) Correlations Regression analysis Distributional analysis Frequency analysis Principal-component analysis Cluster analysis Each form of analysis has strengths and weaknesses CS 640 12

Self-Similar Nature of Network Traffric W. Leland, M. Taqqu, W. Willinger, D. Wilson, On the Self-Similar Nature of Ethernet Traffic, IEEE/ACM TON, 1994. Baker Award winner V. Paxson, S. Floyd, Wide-Area Traffic: The Failure of Poisson Modeling, IEEE/ACM TON, 1995. M. Crovella, A. Bestavros, Self-Similarity in World Wide Web Traffic: Evidence and Possible Causes, IEEE/ACM TON, 1997. CS 640 13

Queuing Models One of the key modeling techniques for computer systems in general Vast literature on queuing theory Nicely suited for network analysis Prof. Mary Vernon is our local expert Generally, queuing systems deal with a situation where jobs (of which there are many) wait in line for a resource (of which there are few) Queuing theory can enable us to determine response time Examples? CS 640 14

Queuing Models contd. Example: packets arriving at a router how can we determine how long it takes for packets to be forwarded by the router? Characteristics necessary to specify a queuing system Arrival process Service time distribution Number of servers System capacity (number of buffers) Population size Service discipline Kendal notation: A/S/m/B/K/SD Response time = waiting time + service time For stability, mean arrival rate must be less than mean service rate CS 640 15

Little s Law One of the most basic theorems in queuing theory (1961) Mean number jobs in system = arrival rate * mean response time Treats a system as a black box Applies whenever number of jobs entering the system equals number of jobs leaving the system No jobs created or lost inside system Can be extended to include systems with finite buffers Example: Average forwarding time in a router is 100 microseconds, I/O rate for packets is 100k. What is the mean number of packets buffered in the router? CS 640 16

Simulation Models Simulation is one of the most common/important methods of analysis/modeling Typically an abstraction of the system under consideration Can provide significant insight to system s behavior Network simulation is difficult because of the different layers of operation and the complexity at each layer Simulation options: build your own, use someone else s Canonical network simulator is ns developed at LBL www.isi.edu/nsnam/ns ssf-net is a new, routing-enabled simulator CS 640 17