具負載平衡之垃圾郵件與病毒郵件防堵系統整合與研究

The system Integration and Researching for Spam-mail and Virus-mail with Load-Balance

陳睿駿(chen ruei jiun)

指導教授 ： 陳士農

亞洲大學/資訊學院/資訊工程學系碩士班/碩士(2007年)

摘要

面對垃圾郵件與病毒郵件日趨氾濫，許多防毒軟體廠商及資安廠商紛紛推出
位於閘道口的硬體過濾產品，其主要目的皆在防堵外來惡意的病毒與垃圾郵件的
攻擊。但這些資安產品大多屬硬體設備，價格動輒百萬或數十萬，成為使用單位
的負擔。因此本研究主要先在UNIX 作業平台架設PostFix 郵件伺服器，再以整
合SpamAssassin 中貝氏過濾法則和ClamAV 防毒軟體，建立垃圾與病毒郵件的
防堵機制，有效地降低建置成本並達成電子郵件收發的控管目標。另外，雖然將
過濾機制以軟體架構設置在伺服器端可減少因大量垃圾郵件造成頻寬負荷以及
降低建置成本等優點，但過濾規則的軟體運算也造成伺服器端系統的龐大負擔，
影響SpamAssassin 的過濾成效。故本研究亦將提出一「具負載平衡之垃圾郵件
與病毒郵件防堵」整合系統，幫助使用者規劃在過濾軟體處理效能的需求基礎之
上，要如何同時搭配硬體最佳化技術來增加過濾垃圾郵件的比率及減少將正常郵
件誤判為垃圾郵件的機率，以滿足不同類別郵件系統在規劃垃圾郵件與病毒郵件
防堵系統的有效架構，以維護網路品質。-

關鍵字

垃圾郵件、病毒郵件、啟發式技術、負載平衡

並列摘要

Face to the spam mail and virus mail problem, a lot of antivirus software
manufacturers and information security solution providers bring up many hardware
filter products, lie in the information gateway and Internet, its main purpose is to
defend the attack of virus and junk mail. But the announced products of information
security mostly belong to the hardware, which tend to be very expensive, become the
burden of the applying unit. This thesis first builds a PostFix mail server in UNIX
platform, and then integrates SpamAssassin and ClamAV to construct spam mail and
virus mail defense mechanism, which to reduce the construction cost and to improve
the effectively.
In addition, although the filter mechanism sets up in server side can reduce
bandwidth usage caused by a large number of spam mails and reduce construction
cost, but the filter rules will cause large workloads in sever system to influence the
performance of SpamAssassin. Therefore, this thesis proposes an integrated system
for spam mail and virus mail with load balance technology, helping users to optimize
efficiency and to reduce error rate. This proposed system can satisfy various mail
systems to improve service quality.

並列關鍵字

Spam Mail 、Virus Mail、 heuristic technology、 load balance.

參考文獻

[2] W .Cohen, Fast effective rule induction. Machine Learning Proceedings of the

Twelfth International Conference. Lake Taho, California, Mongan Kanfmann, pp.

symposium of Machine Learning in Information Access, Palo Alto. California, pp.

18-25. 1996

[4] H. Drucker, D. Wu, and V. N. Vapnik, Support Vector Machines for Spam