Skip to content

Latest commit

 

History

History
40 lines (28 loc) · 1.53 KB

README.md

File metadata and controls

40 lines (28 loc) · 1.53 KB

spark kafka producer consumer example

Build Status

Problem Statement

The customer runs a website and periodically is attacked by a botnet in a Distributed Denial of Service (DDOS) attack. Given a log file in Apache log format from a given attack, use it to build a simple real-time detector of DDOS attacks.

Requirements

  • Ingest
    • Start with reading the file from the HDFS file system
    • Write a tool to read the log file and put messages on the message system such as Kafka
  • Detection
    • Write an application which reads messages from the message system and detects whether the attacker is part of the DDOS attack
    • Once an attacker is found, the ip-address should be written to a results directory which could be used for further processing
    • An attack should be detected one to two minutes after starting

Possible Solution

Workflow Diagram

Assuming a user cannot send multiple requests at a given instance of time, duplicate IP addresses for a timestamp will be marked as suspicious botnet attack. Example - A machine 19.4.31.13 sends request at 25/May/2015:23:11:15 two or more times. This system will detect it as botnet attack and block it from sending any future requests.

Tools

Zookeeper 3.4.5
Kafka 8.2.1
Scala 2.10
Spark 1.6.1
Hadoop 2.6.0
Java 1.7
GGTS 3.6.4.RELEASE
Eclipse Luna SR1 (4.4.2)
Eclipse Buildship Gradle 1.0.16
Eclipse Scala IDE 4.4.1