Whats the Problem Web Server 1 Web Server

What’s the Problem Web Server 1 Web Server N ? ? Which Web Server to Choose? Request Client • Web system played an essential role in Proving and Retrieve information. • Cause Overloaded Status and Longer Response Time • Duplicated Web System is Widely Used Today with Purpose to Scatter the Client’s Request and Shorten the Response Time • Key Problem How to allocate the request efficiently to maximally Shorten the Response Time

Related Works on Web Systems • CISCO Distributed Director… • GIT Feo et al, Dynamic Server Selection, … • NCSA Scalable Web Server….

Web Components and Their Interaction

Two Metrics 1. Current Load L 1, t /S 1 = L 2, t /S 2 = L 3, t /S 3 … = Lk, t /Sk The goal is to Make the Loads Even among the Web Servers 2. Average Response Time (ART) ART load-balancing status = min(ART) The Goal is to Maximally Shorten the Response Time

Factors That Affect Load Balancing • • • Size of Request Web Processing Power The Number of pending Requests at the Web Server The Size of Pending Requests at the Web Server Distance of the Path between Client and Web Server Available Bandwidth along the Path Hop Count of the Path Traffic Status Number of Web Servers Divide into Two Information: Static and Dynamic: Static Information: Web Server Processing Power, Distance, Bandwidth, Web Server Number, Hop Count

Load Balancing Algorithm Design Concern Two Processing Times in Load Balancing Algorithm: 1. Time from Client to Web Server. It Includes Time: Transmission Delay Queuing Delay 2. Time Stay at Web Server. It Includes Time: Queuing at Web Server Processing Time at Web Server Idea Balancing Those Times. Load Balancing Algorithm should try to shorten the End-to-End Response Time of the Request.

Load Balancing Algorithm I 1. LBA-I: Use all Static Information Selected Web Server i = Min{Distance/Bandwidth + Loads of Web Server i / + Process Power of Web Server i} • Use the Ratio of Distance to Bandwidth to Measure the Path, • Use the Ratio of Web Server Loads to Processing Power to Measure the Web Status. There are two Variation Algorithms: LBA-I-1: Add Hop Count LBA-I-2: Just Concern Web Status

Load Balancing Algorithm II LBA-II: LBAs communicate with each other. When a LBA Makes an Assignment Decision, It Passes this info to Other LBAs, Other LBAs Update their Assignment Table. Advantage: Improve the estimated precision of loads of web server. Disadvantage: Generate heavy communication overhead and take away available bandwidth for web access. . A Variation Algorithm LBA-II-1. Just Pass the Assignment info to Neighboring LBAs Advantage: Reduce Communication Overhead. Disadvantage: Decrease the Estimated Precision of Web Server Loads.

Load Balancing Algorithm III Using Static Information to Measure the Path Traffic in LBA-I and LBA-III: Using Dynamic Path Information. Periodically send probing packets to probe path bandwidth. Advantage: Improve the Estimated Precision of Path Traffic. Disadvantage: Generate Heavy Communication Overhead, reduce available bandwidth.

Load Balancing Algorithm IV Using Assignment Number of Request to Measure the Web Server Loads in LBA-I, LBA-II and LBA-III. LBA-IV: Using Left Loads Information to Measure the Web Server Load Status. Web Servers Periodically Send the Left Load Message to LBAs Advantage: Improve the Estimated Precision of Web Server Loads. Disadvantage: 1. Generate Heavy Communication Overhead, reduce available bandwidth. 2. The performance of the algorithm depends on web server reporting period. There are Two Variations of Algorithms LBA-IV(E) LBA-IV(Tc)

Sequence Chart Of Simulation Program

Web Server and Client Traffic Characterizations Log Files of Five Web Server are analyzed: 1. Department wide server at UCCS run on an Alpha workstation owl. uccs. edu. 2. Campus wide server at UCCS run on an Alpha workstation www. uccs. edu. 3. Clark. Net WWW server, which is a full Internet access provider for the Metro Baltimore-Washington DC area. 4. EPA WWW server located at Research Triangle Park, NC. 5. BU-Web-Client in the Boston University Computer Science Department. Characterize the Workload of Web Server and Client. Compare Those Characterization with Other Reports

Characterizations of Document Type Most Frequently Requested by the Clients are Image and HTML type, Which Account Over 80%of all Requested Type.

Characterization of Document Size Most Document Size Requested by Clients is 1 ~ 5 KB

Characterization of Request Time Interval Most Time Interval of Request is less 1 sec

Test Plan There are Following Factors Affecting Performance of Algorithms 1. Network Topology 2. Request Count 3. Request Time Interval 4. Bandwidth We will Change Those Parameters to Test the Performance of the Algorithms. We will collect the statistics of the following Data 1. Average Response Time 2. Web Queuing Delay Time 3. Router Queuing Delay Time 4. Transmission Delay Time 5. Propagation Delay 6. Processing Time 7. Various Communication Overhead

Performance of Load Balancing Algorithms Distribution of Request Time Interval as Figure 6 -2 Distribution of Document Size as Figure 6 -1 Request Count from 800 to 10000 Average Bandwidth as 5 Mb New-Jersey Network as Figure 5 -1 Transmission Delay ( 90%) Dominates the Response Time

Performance of Load Balancing Algorithms Distribution of Request Time Interval as Figure 6 -2 Distribution of Document Size as Figure 6 -1 Request Count from 800 to 10000 Average Bandwidth as 5 Mb r 50 Network as Append A Transmission Delay ( 80%) Dominates the Response Time

Performance of Load Balancing Algorithm From above Figure, we can see when the bandwidth increases, the Average Response Time of all algorithms decrease in the case that the transmission delay dominates the response time of the request

Performance of Load Balancing Algorithms Now We Change the Request Interval Time as Figure 6 -18 (Chow: show Figure 6 -18 here) The Web Queuing Delay Dominates the Response Time of Request. (how much) Following Figure Shows the Performance of Algorithms

Performance of Load Balancing Algorithms

Summary of Load Balancing Algorithm Performance 1. Algorithm LBA-I and LBA-I-1 have the Better Performance When the Transmission Delay Dominates the Response Time. 2. Algorithm LBA-II(I-2) has the Better Performance When the Web Queuing Delay Dominates the Response Time 3. As the Bandwidth Increases, except Algorithm LBA-I and LBA-I-2, all Algorithm Decreases the Response Time. 4. Algorithms LBA-I, LBA-I-1 are Sensitive to Bandwidth Changes

Network Design Issues If the Transmission Delay Dominates the Response Time, We have Following Suggestions for Network Design: 1. Reduce Document Size. 2. Choose Proper Ratio of Web Servers and Clients 3. Choose Proper Process Power of Web Servers 4. Choose Proper Location of Web Servers

Conclusions 1. Algorithm LBA-I and LBA-I-1 have the Better Performance When the Transmission Delay Dominates the Response Time. They are independent of ant Period and Do not Generate any Overhead Messages. 2. Algorithm LBA-II Have the Better Performance When The Web Queuing Delay Dominates the Response Time. It Generates very Heavy Overhead Messages and is independent of reporting Period. 3. Algorithm LBA-III Have the Worse Performance in Both case-Transmission Delay Dominates the Response Time and Web Queuing Delay Dominates Response Time. It Generates Very Heavy Overhead Messages and is dependent on reporting Period. 4. Algorithm LBA-IV has the Better Performance When the Web Queuing Delay Dominates the Response Time. It Generates Overhead Messages and is dependent on reporting period.

Future Directions 1. Using real and larger networks to test the proposed load balancing algorithms 2. Investigate algorithm performance under heavy web server load. 3. Investigate aggregate server/LBA reporting and impact of reporting frequencies. 4. Implement load balancing algorithms in a prototype.