📄 http:^^www.cs.cornell.edu^info^people^vitrano^cs519.html
字号:
MIME-Version: 1.0
Server: CERN/3.0
Date: Sunday, 01-Dec-96 19:31:08 GMT
Content-Type: text/html
Content-Length: 2297
Last-Modified: Thursday, 31-Oct-96 05:28:42 GMT
<html><head><title>Transaction Server - Bart</title></head><body BACKGROUND="stucco.jpg"><body><b><h1 align=center>Transaction Server</h1></b><br><h5 align=center>Libby Lin, Sunil Srivastava, Eric Vitrano</h5><hr><br><h2 align=center>Write-Up: 1</h2><br><h3>Goal:</h3><p><ul>The goal of our project is to create a fault tolerant Transaction Server for multiple clients using TCPIP as the basis for communication. </p></ul><br><h3>Interfaces:</h3><p><ul>We will be providing service to the ECash Application, who will be using our functionstubs, and we will be using the services of Stable Storage, Encryption and possibly Private Email.</p></ul><br><h3>Schedule:</h3><p><ul>The schedule with appropriate methods for achieving the desired goals is as follows :</p></ul><br><center><TABLE BORDER=2 WIDTH=480 CELLPADDING=0 CELLSPACING=0><TR><TD><b>Week</b></TD><TD><b>Goals</b></TD><td><b>Methods</b></td></TR><TR><TD>1.</TD><TD>Single Client and Single Server<br>Fault Tolerance.</TD><td>2 or 3 Phase Commit.<br>TCPIP with encryption layered on top of it.</td></TR><TR><TD>2.</TD><TD>Multiple Client and Single Server</TD><td>Forking on the Server side.</td></TR><TR><TD>3.</TD><TD>Multiple Client and Multiple Server</TD><td>Randomized load balancing on the client's side.<br>i.e. Clients will randomly choose a Transaction Server.</td></TR><TR><TD>4.</TD><TD>Testing/Debugging. *</TD><td></td></TR><TR><TD>5.</TD><TD>Preparation for Presentation. *</TD><td></td></TR></TABLE><br>* If time permits, we may implement a Private Email communication withECash.</center><br><br><h3>Design Decisions:</h3><p><ul>Communication Protocol - TCPIP was chosen to be the communication protocol between ourfunction stubs and transaction servers because SSL is not fully standardized, and may have legal problemswith RSA. Privte Email is our secondary choice because of excessive overhead, but may be implementedlater to provide a higher degree of service.</p></ul><br><h3>References:</h3><p><ul>Dale Skeen and Michael Stonebraker. A Formal Model Crash Recovery in a Distributed System. IEEE Transaction of Software Engineering, May 1983.</p></ul></body></html>⌨️ 快捷键说明
复制代码
Ctrl + C
搜索代码
Ctrl + F
全屏模式
F11
切换主题
Ctrl + Shift + D
显示快捷键
?
增大字号
Ctrl + =
减小字号
Ctrl + -