Building a Cloud Computing Analysis System for  Intrusion Detection System   DATE:4/14/09 Wei-Yu Chen, Yao-Tsung Wang National Center for High-Performance Computing, Taiwan  {waue,jazz}@nchc.org.tw
Taiwan Introduction
NCHC Introduction The NCHC is responsible for Taiwan’s cyberinfrasructure, R&D in HPC and networking applications 3 Business Units located in science parks to support local high tech industry The NCHC integrates information, engineering and scientific disciplines The NCHC provides computing, networking and storage services National Center for High-performance Computing
Outline Motivation The IDEA Architecture Procedure Results Pros and Cons Conclusions
Current Situation of IDS  Intrusion Detection System IDS  (IDS) Detecting unwanted attempts at accessing, manipulating or disabling of computer systems through Internet.  IDS Detect Rate “ false positive”  “ false negative”  Accuracy = top mission of IDS ?
Alerts Alert is produced when IDS detect something as malicious. Two method of alert storage A Text Log -> terrible In Database -> mostly
What’s the problem about Alert ? Enormous Data     less Efficient   Ignore  the crucial information easily  !!! Got  Nothing  if the database were crash
Our Motivation To resolve above problems come with huge amount of anomaly information generated by IDS
Our IDEA - ICAS ICAS,  IDS Cloud Analysis System Applying Cloud Computing technique Improve higher performance of analysis reducing redundancy  Merge relation
System Architecture ICAS Overview
System Architecture SNORT is an open source network intrusion prevention and detection system The most widely deployed intrusion detection Snort
System Architecture Apache Hadoop Core is a software platform that lets one easily write and run applications that process vast amounts of data.  Inspired by Google's MapReduce and Google File System (GFS) papers Implements MapReduce and Hadoop Distributed File System (HDFS)  Operates <key, value> pairs   Hadoop
System Architecture HBase is the Hadoop database  An open-source, distributed, column-oriented store modeled after the Google paper, BigTable HBase
System Architecture Regular Parser Parsing original snort log and transfer to HDFS (hadoop file system) Analysis Procedure  Dispatch job if pool is not empty and  insert the result into database Data Mapper  <key, value> mapping Data Reducer  <“key1”, value1…valueN> <“key2”, value1…valueN> Four Components
Program Procedure
Alert Integration Procedure
Key - Values The victim IP addresses A unique ID used to identify attack method in Snort rules The time when the attack was launghed TCP/IP protocol Attack was lunched from this port Victim ports The IP address where malicious one launghed attack
Alert Merge Example 6007,6008 5002 4077,5002 T5 T4 T1,T2,T3 53 443 80,443 tcp, udp tcp tcp Sip3,Sip4,Sip5 ,Sip6 D.D.O.S. Host_3 Sip1 Trojan Host_2 Sip1,Sip2 Trojan Host_1 Values Key T5 tcp 6008 53 Sip6 D.D.O.S Host_3 T5 udp 6007 53 Sip5 D.D.O.S Host_3 T5 tcp 6008 53 Sip4 D.D.O.S Host_3 T5 udp 6007 53 Sip3 D.D.O.S Host_3 T4 tcp 5002 443 Sip1 Trojan Host_2 T3 tcp 5002 443 Sip1 Trojan Host_1 T2 tcp 4077 80 Sip2 Trojan Host_1 T1 tcp 4077 80 Sip1 Trojan Host_1 Timestamp Packet Protocol Source Port Destination Port Source IP Attack Signature Destination IP
Experiment Environment Machine:  CPU : Intel quad-core, Memory : 2g,   OS :  Linux : Ubuntu 8.04 server Software Hadoop : core 0.16.4 Hbase : 0.1.3 Java : 6 Alerts Data Sets MIT Lincoln Laboratory, Lincoln Lab Data Sets  Computer Security group at UCDavis, tcpdump file
Experimental Result The Consuming Time of Each Number of Data Sets
Experimental Result  Throughput Data Overall
Pros &  Cons Legible Efficient Scalable Economical Reliable Non-realtime Latency immature
Conclusions v.s.  Future Works ICAS supplies a efficient way to analyze and merge huge number of alerts based on cloud platform. Including more IDS logs The best final result is graphical Prepare more large-scale and complete experiment
Thank You ! &  Question ?  DATE:4/14/09

Cloudslam09:Building a Cloud Computing Analysis System for Intrusion Detection

  • 1.
    Building a CloudComputing Analysis System for Intrusion Detection System DATE:4/14/09 Wei-Yu Chen, Yao-Tsung Wang National Center for High-Performance Computing, Taiwan {waue,jazz}@nchc.org.tw
  • 2.
  • 3.
    NCHC Introduction TheNCHC is responsible for Taiwan’s cyberinfrasructure, R&D in HPC and networking applications 3 Business Units located in science parks to support local high tech industry The NCHC integrates information, engineering and scientific disciplines The NCHC provides computing, networking and storage services National Center for High-performance Computing
  • 4.
    Outline Motivation TheIDEA Architecture Procedure Results Pros and Cons Conclusions
  • 5.
    Current Situation ofIDS Intrusion Detection System IDS (IDS) Detecting unwanted attempts at accessing, manipulating or disabling of computer systems through Internet. IDS Detect Rate “ false positive” “ false negative” Accuracy = top mission of IDS ?
  • 6.
    Alerts Alert isproduced when IDS detect something as malicious. Two method of alert storage A Text Log -> terrible In Database -> mostly
  • 7.
    What’s the problemabout Alert ? Enormous Data  less Efficient Ignore the crucial information easily !!! Got Nothing if the database were crash
  • 8.
    Our Motivation Toresolve above problems come with huge amount of anomaly information generated by IDS
  • 9.
    Our IDEA -ICAS ICAS, IDS Cloud Analysis System Applying Cloud Computing technique Improve higher performance of analysis reducing redundancy Merge relation
  • 10.
  • 11.
    System Architecture SNORTis an open source network intrusion prevention and detection system The most widely deployed intrusion detection Snort
  • 12.
    System Architecture ApacheHadoop Core is a software platform that lets one easily write and run applications that process vast amounts of data. Inspired by Google's MapReduce and Google File System (GFS) papers Implements MapReduce and Hadoop Distributed File System (HDFS) Operates <key, value> pairs Hadoop
  • 13.
    System Architecture HBaseis the Hadoop database An open-source, distributed, column-oriented store modeled after the Google paper, BigTable HBase
  • 14.
    System Architecture RegularParser Parsing original snort log and transfer to HDFS (hadoop file system) Analysis Procedure Dispatch job if pool is not empty and insert the result into database Data Mapper <key, value> mapping Data Reducer <“key1”, value1…valueN> <“key2”, value1…valueN> Four Components
  • 15.
  • 16.
  • 17.
    Key - ValuesThe victim IP addresses A unique ID used to identify attack method in Snort rules The time when the attack was launghed TCP/IP protocol Attack was lunched from this port Victim ports The IP address where malicious one launghed attack
  • 18.
    Alert Merge Example6007,6008 5002 4077,5002 T5 T4 T1,T2,T3 53 443 80,443 tcp, udp tcp tcp Sip3,Sip4,Sip5 ,Sip6 D.D.O.S. Host_3 Sip1 Trojan Host_2 Sip1,Sip2 Trojan Host_1 Values Key T5 tcp 6008 53 Sip6 D.D.O.S Host_3 T5 udp 6007 53 Sip5 D.D.O.S Host_3 T5 tcp 6008 53 Sip4 D.D.O.S Host_3 T5 udp 6007 53 Sip3 D.D.O.S Host_3 T4 tcp 5002 443 Sip1 Trojan Host_2 T3 tcp 5002 443 Sip1 Trojan Host_1 T2 tcp 4077 80 Sip2 Trojan Host_1 T1 tcp 4077 80 Sip1 Trojan Host_1 Timestamp Packet Protocol Source Port Destination Port Source IP Attack Signature Destination IP
  • 19.
    Experiment Environment Machine: CPU : Intel quad-core, Memory : 2g, OS : Linux : Ubuntu 8.04 server Software Hadoop : core 0.16.4 Hbase : 0.1.3 Java : 6 Alerts Data Sets MIT Lincoln Laboratory, Lincoln Lab Data Sets Computer Security group at UCDavis, tcpdump file
  • 20.
    Experimental Result TheConsuming Time of Each Number of Data Sets
  • 21.
    Experimental Result Throughput Data Overall
  • 22.
    Pros & Cons Legible Efficient Scalable Economical Reliable Non-realtime Latency immature
  • 23.
    Conclusions v.s. Future Works ICAS supplies a efficient way to analyze and merge huge number of alerts based on cloud platform. Including more IDS logs The best final result is graphical Prepare more large-scale and complete experiment
  • 24.
    Thank You !& Question ? DATE:4/14/09

Editor's Notes

  • #2 Hello every body, I am wei-yu chen, the other one is yao-tsung wang. We are umpired by nchc in beautiful Taiwan. I am very glad to present the talk,”Building a cloud computing analysis system for instrusion detection system”, that is mention about using cloud computing technique to solve security issue. ? Because of this session is the last session, I would introduce this talk as soon as possible.