Analysis and Evaluation of the Windows Event Log - Bill Buchanan

More documents

Recommendations

Info

Barrie Codona, BSc (Hons) Network Computing, 2007 6 Evaluation 6.1 Introduction Initially the application is tested to identify the correctness and completeness of the developed software, any faults or problems that are identified will be dealt with at the maintenance stage. After modifications have been made and documented, the application will then undergo further testing that will measure the burden that it places on the system. The application will be stress tested; a large number of events will be generated to check if the application is able to detect all of them. An attempt will also be made to see if it is possible to try and calculate the HMAC key using a brute force algorithm. Based upon the results of this test it will be possible to recommend a suitable sized key. 6.2 Initial Testing This phase was to test the initial design that had been implemented. Its purpose was to identify if the Event Logging Application could successfully capture a large amount of information without missing any events and to identify any points in the system that needed to be modified beyond their original design specification. The testing was carried out using the automatic testing application that had been developed. Initially 1,000 events were generated and captured, it had been noticed that there was a delay in getting all these events written to file by the Server. The same test was then carried out several more times with the number of events increasing by 1,000 each time. One point to note is that the application became unresponsive when 5,000 events were generated. After the Initial Testing phase was completed a serious flaw in the application was detected, namely a large number of events being generated caused a buffer overflow in the .Net FileSystemWatcher class. The MSDN website says that “If there are many changes in a short time, the buffer can overflow. This causes the component to lose track of changes in the directory, and it will only provide blanket notification.” 51
Barrie Codona, BSc (Hons) Network Computing, 2007 (Microsoft, 2008). This caused the FileSystemWatcher to crash and it then failed to capture any other events, until the application was restarted. It had been suspected that the buffer overflow problem may have been caused by either the size of the buffer, the HMAC tagging or the asymmetric encryption that was happening after the events were captured; Figure 34 explains the process. Figure 34: Data flow Performance tests were then carried out on each of the individual components; this initially consisted of increasing the size of the buffer, which made very little difference when dealing with upwards of 5,000 events. A test was created that would have the FileSystemWatcher capture events and send them directly to the TCP Stream, thus bypassing the HMAC and Asymmetric Encryption. It had then been noted that the previous buffer overflow problem had gone, even when the buffer size was set back to its default value of 8K it still managed to successfully capture 20,000 events. After this successful test the HMAC tagging was added back in, this too did not hinder the performance of the application, which left only the asymmetric encryption process to be responsible. 6.3 Maintenance Based upon the preliminary set of test results, the Event Logging Client and Server applications were heavily modified to improve the encryption process. This involved the use of symmetric encryption for handling the transfer of the events from the Client to the Server. The shared key that the symmetric encryption uses is transferred from the Client to the Server using asymmetric encryption. This brings in several new stages for the Client Server Communication Protocol that was originally described in the Design chapter. Figure 35 explains the new process: 52
Page 1 and 2: Analysis and Evaluation of the Wind
Page 3 and 4: Barrie Codona, BSc (Hons) Network C
Page 51: Barrie Codona, BSc (Hons) Network C
Page 75 and 76: Appendix A: Diary Sheets
Page 77 and 78: NAPIER UNIVERSITY SCHOOL OF COMPUTI
Page 85 and 86: Appendix B: Preliminary Gantt Chart
Page 87 and 88: Nov '07 Dec '07 Jan '08 Feb '08 Mar
Page 89 and 90: C:\Documents and Settings\Barrie\My
Page 95 and 96: Appendix D: Server Application Code
Page 99 and 100: Appendix E: Event Viewer Code
Page 103 and 104:
C:\Documents and Settings\Barrie\My
Page 105 and 106:
Appendix G: Processor Monitor Code
Page 107 and 108:
Appendix H: HMAC Brute Force Cracke
Page 109 and 110:
Page 111 and 112:
Page 113 and 114:
Appendix I: Windows Event Log Tests
Page 115 and 116:
Copy Security Log C:\windows\system
Page 117 and 118:
Another look at the Security Log re
Page 119 and 120:
The Event Service is then restarted
Page 121 and 122:
Another look at the ‘c:\windows\s
Page 123 and 124:
CO42019 - Project 4 Computer Name P
Page 125 and 126:
CO42019 - Project 4 Turned off the
Page 127 and 128:
CO42019 - Project 4 And then restar
Page 129 and 130:
CO42019 - Project 4 Modify the ‘l
Page 131 and 132:
CO42019 - Project 4 Project - Week
Page 133 and 134:
Page 135 and 136:
Page 137 and 138:
CO42019 - Project 4 As previously d
Page 139 and 140:
CO42019 - Project 4 The screenshot
Page 141 and 142:
CO42019 - Project 4 Displaying the
Page 143 and 144:
CO42019 - Project 4 Modifying the r
Page 145 and 146:
CO42019 - Project 4 Master File Tab
Page 147 and 148:
CO42019 - Project 4 Modify the sect
Page 149 and 150:
CO42019 - Project 4 Sectors 3748520
Page 151 and 152:
CO42019 - Project 4 This now shows
show all

Analysis and Evaluation of the Windows Event Log - Bill Buchanan

Create successful ePaper yourself

Delete template?

Save as template?