Download presentation
Presentation is loading. Please wait.
Published bySydnee Bracey Modified over 9 years ago
1
ARC: Protecting against HTTP Parameter Pollution Attacks Using Application Request Caches Elias Athanasopoulos, Vassileios P. Kemerlis, Michalis Polychronakis Columbia University (US) Evangelos P. Markatos FORTH-ICS (GREECE) ACNS 2012
2
Web Applications ARC, ACNS 2012Elias Athanasopoulos, Columbia University2 Web Server Web Browser HTTP Request GET login?username=joe HTTP Request GET login?username=joe HTTP Response HTTP OK
3
URLs in HTTP URL: login?username=joe Action: login Parameters: username ARC, ACNS 2012Elias Athanasopoulos, Columbia University3 HTTP Request GET login?username=joe HTTP Request GET login?username=joe
4
Example: Web e-mail ARC, ACNS 2012Elias Athanasopoulos, Columbia University4 Login Read e-mail Delete e-mail Delete mailbox Logout login?username=joe action?type=read&id=42 action?type=delete&id=42 action?type=del_box&id=inbox logout?username=joe
5
Are all URLs valid? ARC, ACNS 2012Elias Athanasopoulos, Columbia University5 login?username=joe&type=delete&id=42 action?type=read&id=42&id=2 action?type=delete&id=2&id=42 action?type=del_box logout?username=joe&type=del_mbox&id=inbox
6
HTTP Parameter Pollution (HPP) How is this URL interpreted? Parsing goes from left to right (6 wins) Parsing goes from right to left (42 wins) Parsing direction does not matter (6 and 42, or 42 and 6 are concatenated) ARC, ACNS 2012Elias Athanasopoulos, Columbia University6 action?type=read&id=6&id=42 action?type=read&id=6
7
ARC, ACNS 2012Elias Athanasopoulos, Columbia University7
8
URL Space ARC, ACNS 2012Elias Athanasopoulos, Columbia University8 All possible URLs URLs that define Web Application’s Logic Attacker URLs
9
HPP Impact About 1,499 of 5,000 highly ranked in Alexa.com web sites are considered vulnerable to HPP exploitation Automated discovery of parameter pollution vulnerabilities in web applications. Balduzzi et al., NDSS 2011. ARC, ACNS 2012Elias Athanasopoulos, Columbia University9
10
ARC, ACNS 2012Elias Athanasopoulos, Columbia University10
11
Goal ARC, ACNS 2012Elias Athanasopoulos, Columbia University11 All possible URLs URLs that define Web Application’s Logic Attacker URLs We need to serve these We need to block these
12
URL Schema A URL schema has the form of: action?par1=&par2=...&parN= ARC, ACNS 2012Elias Athanasopoulos, Columbia University12 login=?username=joelogin?username=
13
Architecture Training phase Deployment phase ARC, ACNS 2012Elias Athanasopoulos, Columbia University13 Web application Passive Monitoring Collection of Legitimate URL schemas Web application ARC Client HTTP Request
14
Training Phase Large frameworks (such as phpBB) are developed and tested by a large community Big applications (like Facebook) test new features in a close environment ARC, ACNS 2012Elias Athanasopoulos, Columbia University14
15
ARC at run-time ARC, ACNS 2012Elias Athanasopoulos, Columbia University15 Web application ARC URL Schemas action?par1=&par2=&…&parN= HTTP Request Valid Schema Exists No Schema Reject Request
16
Implementation ARC is a web application proxy implemented in Google’s Go ARC uses Go structures for hash tables and lists, Go channels for multithreading ARC, ACNS 2012Elias Athanasopoulos, Columbia University16
17
Data Structures ARC, ACNS 2012Elias Athanasopoulos, Columbia University17 action type=id= action?type=forward&id=42&to=mark to=
18
ARC, ACNS 2012Elias Athanasopoulos, Columbia University18
19
Synthetic Traces Web AppURLsMin Par.Max Par.Density Small1,0005120.01 Medium10,0007150.001 Heavy100,00012200.001 ARC, ACNS 2012Elias Athanasopoulos, Columbia University19 Density: ratio of unique actions over all possible URL schemas.
20
Trace Selection ARC, ACNS 2012Elias Athanasopoulos, Columbia University20
21
Multithreading We have implemented two version of ARC Single Channel 4-Channel ARC, ACNS 2012Elias Athanasopoulos, Columbia University21
22
Request Resolution ARC, ACNS 2012Elias Athanasopoulos, Columbia University22 Requests are resolved in less than 10 microseconds.
23
Throughput ARC, ACNS 2012Elias Athanasopoulos, Columbia University23 Requests can be processed in a rate of hundreds of thousand URLS per second
24
Takeaways ARC can protect HPP vulnerable applications by keeping a white list of accepted URL schemas ARC is fast and can be transparently applied to legacy web applications ARC, ACNS 2012Elias Athanasopoulos, Columbia University24
25
ARC, ACNS 2012Elias Athanasopoulos, Columbia University25
26
HTTP Parameter Pollution (HPP) New attack targeting web applications HTTP parameters injection Manipulation of web application’s control flow Drive a web application according to attacker’s needs ARC, ACNS 2012Elias Athanasopoulos, Columbia University26
27
HPP in a slide Web applications are driven through HTTP requests and responses, which encapsulate resource descriptors: URLs URLs are composed by an action and a list of parameters http://site/login?user=joe&country=SG The list of parameters can be polluted with extra parameters ARC, ACNS 2012Elias Athanasopoulos, Columbia University27
28
URL example This URL is associated with a script purchase, which is called with input argument item_id which has the value 42 http://www.e-store.com/purchase?item_id=42 ARC, ACNS 2012Elias Athanasopoulos, Columbia University28 Action: purchase Parameter: item_id=42
29
Attack Scenario: e-store Two families of URLs: (1) show?category=1 (2) purchase?category=1&item_id=1 ARC, ACNS 2012Elias Athanasopoulos, Columbia University29
30
Normal Operation ARC, ACNS 2012Elias Athanasopoulos, Columbia University30 show?category=1 purchase?item_id=1 purchase?item_id=2 … purchase?item_id=N purchase?item_id=1 purchase?item_id=2 … purchase?item_id=N category=1 + + purchase?item_id=1&category=1 purchase?item_id=2&category=1 … purchase?item_id=N&category=1 purchase?item_id=1&category=1 purchase?item_id=2&category=1 … purchase?item_id=N&category=1
31
Bob attacks Bob lures Alice to click on links, like: show?category=1%26item_id=42 Channels: IM, fake web pages, e-mail, etc. ARC, ACNS 2012Elias Athanasopoulos, Columbia University31
32
HPP in Action ARC, ACNS 2012Elias Athanasopoulos, Columbia University32 show?category= 1%26item_id=42 purchase?item_id=1 purchase?item_id=2 … purchase?item_id=N purchase?item_id=1 purchase?item_id=2 … purchase?item_id=N category=1 %26item_id=42 + + purchase?item_id=1&category=1&item_id=42 purchase?item_id=2&category=1&item_id=42 … purchase?item_id=N&category=1&item_id=42 purchase?item_id=1&category=1&item_id=42 purchase?item_id=2&category=1&item_id=42 … purchase?item_id=N&category=1&item_id=42
33
Normal Operation vs HPP ARC, ACNS 2012Elias Athanasopoulos, Columbia University33 purchase?item_id=1&category=1&item_id=42 purchase?item_id=1&category=1
34
ARC, ACNS 2012Elias Athanasopoulos, Columbia University34
Similar presentations
© 2025 SlidePlayer.com. Inc.
All rights reserved.