This site is the archived OWASP Foundation Wiki and is no longer accepting Account Requests.
To view the new OWASP Foundation website, please visit https://owasp.org

Difference between revisions of "Map Application Architecture (OTG-INFO-010)"

From OWASP
Jump to: navigation, search
(Map and Test phases divided)
(Final edit)
Line 2: Line 2:
  
 
== Brief Summary ==
 
== Brief Summary ==
The intrinsic complexity of interconnected and heterogeneous web server infrastructure, which can count hundreds of web applications, makes configuration management and review a fundamental step in testing and deploying every single application.
+
The intrinsic complexity of interconnected and heterogeneous web server infrastructure, which can include hundreds of web applications, makes configuration management and review a fundamental step in testing and deploying every single application.
In fact it takes only a single vulnerability to undermine the security of the entire infrastructure, and even small and (almost) unimportant problems may evolve into severe risks for another application on the same server.
+
In fact it takes only a single vulnerability to undermine the security of the entire infrastructure, and even small and (almost) unimportant problems may evolve into severe risks for another application on the same server.  
In order to address these problems, it is of utmost importance to perform an in-depth review of configuration and known security issues.
+
 
Before performing an in-depth review it is necessary to map the network and application architecture.
+
 
 +
To address these problems, it is of utmost importance to perform an in-depth review of configuration and known security issues. Before performing an in-depth review it is necessary to map the network and application architecture.
 +
 
  
 
== Description of the Issue ==  
 
== Description of the Issue ==  
  
 
Proper configuration management of the web server infrastructure is very important in order to preserve the security of the application itself. If elements such as the web server software, the back-end database servers, or the authentication servers are not properly reviewed and secured, they might introduce undesired risks or introduce new vulnerabilities that might compromise the application itself.
 
Proper configuration management of the web server infrastructure is very important in order to preserve the security of the application itself. If elements such as the web server software, the back-end database servers, or the authentication servers are not properly reviewed and secured, they might introduce undesired risks or introduce new vulnerabilities that might compromise the application itself.
 +
  
 
For example, a web server vulnerability that would allow a remote attacker to disclose the source code of the application itself (a vulnerability that has arisen a number of times in both web servers or application servers) could compromise the application, as anonymous users could use the information disclosed in the source code to leverage attacks against the application or its users.
 
For example, a web server vulnerability that would allow a remote attacker to disclose the source code of the application itself (a vulnerability that has arisen a number of times in both web servers or application servers) could compromise the application, as anonymous users could use the information disclosed in the source code to leverage attacks against the application or its users.
 +
  
 
In order to test the configuration management infrastructure, the following steps need to be taken:
 
In order to test the configuration management infrastructure, the following steps need to be taken:
  
* The different elements that make up the infrastructure need to be determined in order to understand how they interact with a web application and how they affect its security.
+
* The different elements that make up the infrastructure need to be determined to understand how they interact with a web application and how they affect security.
* All the elements of the infrastructure need to be reviewed in order to make sure that they don’t hold any known vulnerabilities.
+
* All the elements of the infrastructure need to be reviewed to make sure that they don’t hold any known vulnerabilities.
 
* A review needs to be made of the administrative tools used to maintain all the different elements.
 
* A review needs to be made of the administrative tools used to maintain all the different elements.
 
* The authentication systems, if any, need to reviewed in order to assure that they serve the needs of the application and that they cannot be manipulated by external users to leverage access.
 
* The authentication systems, if any, need to reviewed in order to assure that they serve the needs of the application and that they cannot be manipulated by external users to leverage access.
 
* A list of defined ports which are required for the application should be maintained and kept under change control.
 
* A list of defined ports which are required for the application should be maintained and kept under change control.
  
In this chapter it is going to be shown how it is possible to map different elements that make up the infrastructure.
+
 
 +
This section shows how to map different elements that make up the infrastructure.
 +
 
  
 
== Black Box Testing and examples==
 
== Black Box Testing and examples==
Line 27: Line 33:
 
===Map the application architecture===
 
===Map the application architecture===
  
The application architecture needs to be mapped through some test to determine what different components are used to build the web application. In small setups, such as a simple CGI-based application, a single server might be used that runs the web server which executes the C, Perl, or Shell CGIs application, and perhaps also the authentication mechanism. On more complex setups, such as an online bank system, multiple servers might be involved including: a reverse proxy, a front-end web server, an application server and a database server or LDAP server. Each of these servers will be used for different purposes and might be even be divided in different networks with firewalling devices between them, creating different DMZs so that access to the web server will not grant a remote user access to the authentication mechanism itself, and so that compromises of the different elements of the architecture can be isolated in a way such that they will not compromise the whole architecture.
+
The application architecture needs to be mapped through some test to determine what different components are used to build the web application. In small setups, such as a simple CGI-based application, a single server might be used that runs the web server which executes the C, Perl, or Shell CGIs application, and perhaps also the authentication mechanism.  
 +
 
 +
 
 +
On more complex setups, such as an online bank system, multiple servers might be involved. These may include a reverse proxy, a front-end web server, an application server and a database server or LDAP server. Each of these servers will be used for different purposes and might be even be divided in different networks with firewalls between them. This creates different DMZs so that access to the web server will not grant a remote user access to the authentication mechanism itself, and so that compromises of the different elements of the architecture can be isolated so that they will not compromise the whole architecture.
 +
 
  
 
Getting knowledge of the application architecture can be easy if this information is provided to the testing team by the application developers in document form or through interviews, but can also prove to be very difficult if doing a blind penetration test.
 
Getting knowledge of the application architecture can be easy if this information is provided to the testing team by the application developers in document form or through interviews, but can also prove to be very difficult if doing a blind penetration test.
  
In the latter case, a tester will first start with the assumption that there is a simple setup (a single server) and will, through the information retrieved from other tests, derive the different elements and question this assumption that the architecture will be extended. The tester will start by asking simple questions such as: “Is there a firewalling system protecting the web server?” which will be answered based on the results of network scans targeted at the web server and the analysis of whether the network ports of the web server are being filtered in the network edge (no answer or ICMP unreachables are received) or if the server is directly connected to the Internet (i.e. returns RST packets for all non-listening ports). This analysis can be enhanced in order to determine the type of firewall system used based on network packet tests: is it a stateful firewall or is it an access list filter on a router? How is it configured? Can it be bypassed?
 
  
Detecting a reverse proxy in front of the web server needs to be done by the analysis of the web server banner, which might directly disclose the existence of a reverse proxy (for example, if ‘WebSEAL’[1]  is returned). It can also be determined by obtaining the answers given by the web server to requests and comparing them to the expected answers. For example, some reverse proxies act as “intrusion prevention systems” (or web-shields) by blocking known attacks targeted at the web server. If the web server is known to answer with a 404 message to a request which targets an unavailable page and returns a different error message for some common web attacks like those done by CGI scanners, it might be an indication of a reverse proxy (or an application-level firewall) which is filtering the requests and returning a different error page than the one expected. Another example: if the web server returns a set of available HTTP methods (including TRACE) but the expected methods return errors then there is probably something in between, blocking them. In some cases, even the protection system gives itself away:
+
In the latter case, a tester will first start with the assumption that there is a simple setup (a single server). Then they will retrieve information from other tests and derive the different elements, question this assumption and extend the architecture map. The tester will start by asking simple questions such as: “Is there a firewalling system protecting the web server?”. This question will be answered based on the results of network scans targeted at the web server and the analysis of whether the network ports of the web server are being filtered in the network edge (no answer or ICMP unreachables are received) or if the server is directly connected to the Internet (i.e. returns RST packets for all non-listening ports). This analysis can be enhanced to determine the type of firewall used based on network packet tests. Is it a stateful firewall or is it an access list filter on a router? How is it configured? Can it be bypassed?
 +
 
 +
 
 +
Detecting a reverse proxy in front of the web server needs to be done by the analysis of the web server banner, which might directly disclose the existence of a reverse proxy (for example, if ‘WebSEAL’[1]  is returned). It can also be determined by obtaining the answers given by the web server to requests and comparing them to the expected answers. For example, some reverse proxies act as “intrusion prevention systems” (or web-shields) by blocking known attacks targeted at the web server. If the web server is known to answer with a 404 message to a request that targets an unavailable page and returns a different error message for some common web attacks like those done by CGI scanners, it might be an indication of a reverse proxy (or an application-level firewall) which is filtering the requests and returning a different error page than the one expected. Another example: if the web server returns a set of available HTTP methods (including TRACE) but the expected methods return errors then there is probably something in between blocking them.  
 +
 
 +
In some cases, even the protection system gives itself away:
  
 
<pre>
 
<pre>
Line 49: Line 63:
 
FW-1 at XXXXXX: Access denied.</BODY>
 
FW-1 at XXXXXX: Access denied.</BODY>
 
</pre>
 
</pre>
 +
  
 
'''Example of the security server of Check Point Firewall-1 NG AI “protecting” a web server'''
 
'''Example of the security server of Check Point Firewall-1 NG AI “protecting” a web server'''
  
Reverse proxies can also be introduced as proxy-caches to accelerate the performance of back-end application servers. Detecting these proxies can be done based, again, on the server header or by timing requests that should be cached by the server and comparing the time taken to server the first request with subsequent requests.
+
Reverse proxies can also be introduced as proxy-caches to accelerate the performance of back-end application servers. Detecting these proxies can be done based on the server header. They can also be detected by timing requests that should be cached by the server and comparing the time taken to server the first request with subsequent requests.
 +
 
 +
 
 +
Another element that can be detected is network load balancers. Typically, these systems will balance a given TCP/IP port to multiple servers based on different algorithms (round-robin, web server load, number of requests, etc.). Thus, the detection of this architecture element needs to be done by examining multiple requests and comparing results to determine if the requests are going to the same or different web servers. For example, based on the Date header if the server clocks are not synchronized. In some cases, the network load balance process might inject new information in the headers that will make it stand out distinctively, like the AlteonP cookie introduced by Nortel’s Alteon WebSystems load balancer.
  
Another element that can be detected: network load balancers. Typically, these systems will balance a given TCP/IP port to multiple servers based on different algorithms (round-robin, web server load, number of requests, etc.). Thus, the detection of this architecture element needs to be done by examining multiple requests and comparing results in order to determine if the requests are going to the same or different web servers. For example, based on the Date: header if the server clocks are not synchronized. In some cases, the network load balance process might inject new information in the headers that will make it stand out distinctively, like the AlteonP cookie introduced by Nortel’s Alteon WebSystems load balancer.
 
  
 
Application web servers are usually easy to detect. The request for several resources is handled by the application server itself (not the web server) and the response header will vary significantly (including different or additional values in the answer header). Another way to detect these is to see if the web server tries to set cookies which are indicative of an application web server being used (such as the JSESSIONID provided by some J2EE servers), or to rewrite URLs automatically to do session tracking.
 
Application web servers are usually easy to detect. The request for several resources is handled by the application server itself (not the web server) and the response header will vary significantly (including different or additional values in the answer header). Another way to detect these is to see if the web server tries to set cookies which are indicative of an application web server being used (such as the JSESSIONID provided by some J2EE servers), or to rewrite URLs automatically to do session tracking.
  
Authentication backends (such as LDAP directories, relational databases, or RADIUS servers) however, are not as easy to detect from an external point of view in an immediate way, since they will be hidden by the application itself.
 
  
The use of a database backend can be determined simply by navigating an application. If there is highly dynamic content generated “on the fly," it is probably being extracted from some sort of database by the application itself. Sometimes the way information is requested might give insight to the existence of a database back-end. For example, an online shopping application that uses numeric identifiers (‘id’) when browsing the different articles in the shop.  However, when doing a blind application test, knowledge of the underlying database is usually only available when a vulnerability surfaces in the application, such as poor exception handling or susceptibility to SQL injection.
+
Authentication back ends (such as LDAP directories, relational databases, or RADIUS servers) however, are not as easy to detect from an external point of view in an immediate way, since they will be hidden by the application itself.
 +
 
 +
 
 +
The use of a back end database can be determined simply by navigating an application. If there is highly dynamic content generated “on the fly," it is probably being extracted from some sort of database by the application itself. Sometimes the way information is requested might give insight to the existence of a database back-end. For example, an online shopping application that uses numeric identifiers (‘id’) when browsing the different articles in the shop.  However, when doing a blind application test, knowledge of the underlying database is usually only available when a vulnerability surfaces in the application, such as poor exception handling or susceptibility to SQL injection.
 +
 
  
 
===Administrative tools===
 
===Administrative tools===
  
Any web server infrastructure requires the existence of administrative tools to maintain and update the information used by the application: static content (web pages, graphic files), application source code, user authentication databases, etc. Depending on the site, technology, or software used, administrative tools will differ. For example, some web servers will be managed using administrative interfaces which are, themselves, web servers (such as the iPlanet web server) or will be administrated by plain text configuration files (in the Apache case[2]) or use operating-system GUI tools (when using Microsoft’s IIS server or ASP.Net). In most cases, however, the server configuration will be handled using different file maintenance tools used by the web server, which are managed through FTP servers, WebDAV, network file systems (NFS, CIFS) or other mechanisms. Obviously, the operating system of the elements that make up the application architecture will also be managed using other tools. Applications may also have administrative interfaces embedded in them that are used to manage the application data itself (users, content, etc.).
+
Any web server infrastructure requires the existence of administrative tools to maintain and update the information used by the application. This information can include static content (web pages, graphic files), application source code, user authentication databases, etc. Depending on the site, technology, or software used, administrative tools will differ. For example, some web servers will be managed using administrative interfaces which are, themselves, web servers (such as the iPlanet web server) or will be administrated by plain text configuration files (in the Apache case[2]) or use operating-system GUI tools (when using Microsoft’s IIS server or ASP.Net).  
 +
 
 +
 
 +
In most cases, however, the server configuration will be handled using different file maintenance tools used by the web server, which are managed through FTP servers, WebDAV, network file systems (NFS, CIFS) or other mechanisms. Obviously, the operating system of the elements that make up the application architecture will also be managed using other tools. Applications may also have administrative interfaces embedded in them that are used to manage the application data itself (users, content, etc.).
 +
 
  
Map of the administrative interfaces used to manage the different parts of the architecture is very important, since if an attacker gains access to any of them he can then compromise or damage the application architecture. Thus in this phase it is important to:
+
Creating a map of the administrative interfaces used to manage the different parts of the architecture is very important, since if an attacker gains access to any of them he can then compromise or damage the application architecture. Thus in this phase it is important to:
  
 
* List all the possible administrative interfaces.
 
* List all the possible administrative interfaces.
 
* Determine if administrative interfaces are available from an internal network or are also available from the Internet.
 
* Determine if administrative interfaces are available from an internal network or are also available from the Internet.
 +
  
 
Some companies choose not to manage all aspects of their web server applications, but may have other parties managing the content delivered by the web application. This external company might either provide only parts of the content (news updates or promotions) or might manage the web server completely (including content and code). It is common to find administrative interfaces available from the Internet in these situations, since using the Internet is cheaper than providing a dedicated line that will connect the external company to the application infrastructure through a management-only interface. In this situation, it is very important to test if the administrative interfaces can be vulnerable to attacks.
 
Some companies choose not to manage all aspects of their web server applications, but may have other parties managing the content delivered by the web application. This external company might either provide only parts of the content (news updates or promotions) or might manage the web server completely (including content and code). It is common to find administrative interfaces available from the Internet in these situations, since using the Internet is cheaper than providing a dedicated line that will connect the external company to the application infrastructure through a management-only interface. In this situation, it is very important to test if the administrative interfaces can be vulnerable to attacks.
 +
  
 
==References==
 
==References==
 
* [1] WebSEAL, also known as Tivoli Authentication Manager, is a reverse proxy from IBM which is part of the Tivoli framework.
 
* [1] WebSEAL, also known as Tivoli Authentication Manager, is a reverse proxy from IBM which is part of the Tivoli framework.
 
* [2] There are some GUI-based administration tools for Apache (like NetLoony) but they are not in widespread use yet.
 
* [2] There are some GUI-based administration tools for Apache (like NetLoony) but they are not in widespread use yet.

Revision as of 12:25, 13 May 2014

This article is part of the new OWASP Testing Guide v4.
Back to the OWASP Testing Guide v4 ToC: https://www.owasp.org/index.php/OWASP_Testing_Guide_v4_Table_of_Contents Back to the OWASP Testing Guide Project: https://www.owasp.org/index.php/OWASP_Testing_Project

Brief Summary

The intrinsic complexity of interconnected and heterogeneous web server infrastructure, which can include hundreds of web applications, makes configuration management and review a fundamental step in testing and deploying every single application. In fact it takes only a single vulnerability to undermine the security of the entire infrastructure, and even small and (almost) unimportant problems may evolve into severe risks for another application on the same server.


To address these problems, it is of utmost importance to perform an in-depth review of configuration and known security issues. Before performing an in-depth review it is necessary to map the network and application architecture.


Description of the Issue

Proper configuration management of the web server infrastructure is very important in order to preserve the security of the application itself. If elements such as the web server software, the back-end database servers, or the authentication servers are not properly reviewed and secured, they might introduce undesired risks or introduce new vulnerabilities that might compromise the application itself.


For example, a web server vulnerability that would allow a remote attacker to disclose the source code of the application itself (a vulnerability that has arisen a number of times in both web servers or application servers) could compromise the application, as anonymous users could use the information disclosed in the source code to leverage attacks against the application or its users.


In order to test the configuration management infrastructure, the following steps need to be taken:

  • The different elements that make up the infrastructure need to be determined to understand how they interact with a web application and how they affect security.
  • All the elements of the infrastructure need to be reviewed to make sure that they don’t hold any known vulnerabilities.
  • A review needs to be made of the administrative tools used to maintain all the different elements.
  • The authentication systems, if any, need to reviewed in order to assure that they serve the needs of the application and that they cannot be manipulated by external users to leverage access.
  • A list of defined ports which are required for the application should be maintained and kept under change control.


This section shows how to map different elements that make up the infrastructure.


Black Box Testing and examples

Map the application architecture

The application architecture needs to be mapped through some test to determine what different components are used to build the web application. In small setups, such as a simple CGI-based application, a single server might be used that runs the web server which executes the C, Perl, or Shell CGIs application, and perhaps also the authentication mechanism.


On more complex setups, such as an online bank system, multiple servers might be involved. These may include a reverse proxy, a front-end web server, an application server and a database server or LDAP server. Each of these servers will be used for different purposes and might be even be divided in different networks with firewalls between them. This creates different DMZs so that access to the web server will not grant a remote user access to the authentication mechanism itself, and so that compromises of the different elements of the architecture can be isolated so that they will not compromise the whole architecture.


Getting knowledge of the application architecture can be easy if this information is provided to the testing team by the application developers in document form or through interviews, but can also prove to be very difficult if doing a blind penetration test.


In the latter case, a tester will first start with the assumption that there is a simple setup (a single server). Then they will retrieve information from other tests and derive the different elements, question this assumption and extend the architecture map. The tester will start by asking simple questions such as: “Is there a firewalling system protecting the web server?”. This question will be answered based on the results of network scans targeted at the web server and the analysis of whether the network ports of the web server are being filtered in the network edge (no answer or ICMP unreachables are received) or if the server is directly connected to the Internet (i.e. returns RST packets for all non-listening ports). This analysis can be enhanced to determine the type of firewall used based on network packet tests. Is it a stateful firewall or is it an access list filter on a router? How is it configured? Can it be bypassed?


Detecting a reverse proxy in front of the web server needs to be done by the analysis of the web server banner, which might directly disclose the existence of a reverse proxy (for example, if ‘WebSEAL’[1] is returned). It can also be determined by obtaining the answers given by the web server to requests and comparing them to the expected answers. For example, some reverse proxies act as “intrusion prevention systems” (or web-shields) by blocking known attacks targeted at the web server. If the web server is known to answer with a 404 message to a request that targets an unavailable page and returns a different error message for some common web attacks like those done by CGI scanners, it might be an indication of a reverse proxy (or an application-level firewall) which is filtering the requests and returning a different error page than the one expected. Another example: if the web server returns a set of available HTTP methods (including TRACE) but the expected methods return errors then there is probably something in between blocking them.

In some cases, even the protection system gives itself away:

GET /web-console/ServerInfo.jsp%00 HTTP/1.0

HTTP/1.0 200
Pragma: no-cache
Cache-Control: no-cache
Content-Type: text/html
Content-Length: 83

<TITLE>Error</TITLE>
<BODY>
<H1>Error</H1>
FW-1 at XXXXXX: Access denied.</BODY>


Example of the security server of Check Point Firewall-1 NG AI “protecting” a web server

Reverse proxies can also be introduced as proxy-caches to accelerate the performance of back-end application servers. Detecting these proxies can be done based on the server header. They can also be detected by timing requests that should be cached by the server and comparing the time taken to server the first request with subsequent requests.


Another element that can be detected is network load balancers. Typically, these systems will balance a given TCP/IP port to multiple servers based on different algorithms (round-robin, web server load, number of requests, etc.). Thus, the detection of this architecture element needs to be done by examining multiple requests and comparing results to determine if the requests are going to the same or different web servers. For example, based on the Date header if the server clocks are not synchronized. In some cases, the network load balance process might inject new information in the headers that will make it stand out distinctively, like the AlteonP cookie introduced by Nortel’s Alteon WebSystems load balancer.


Application web servers are usually easy to detect. The request for several resources is handled by the application server itself (not the web server) and the response header will vary significantly (including different or additional values in the answer header). Another way to detect these is to see if the web server tries to set cookies which are indicative of an application web server being used (such as the JSESSIONID provided by some J2EE servers), or to rewrite URLs automatically to do session tracking.


Authentication back ends (such as LDAP directories, relational databases, or RADIUS servers) however, are not as easy to detect from an external point of view in an immediate way, since they will be hidden by the application itself.


The use of a back end database can be determined simply by navigating an application. If there is highly dynamic content generated “on the fly," it is probably being extracted from some sort of database by the application itself. Sometimes the way information is requested might give insight to the existence of a database back-end. For example, an online shopping application that uses numeric identifiers (‘id’) when browsing the different articles in the shop. However, when doing a blind application test, knowledge of the underlying database is usually only available when a vulnerability surfaces in the application, such as poor exception handling or susceptibility to SQL injection.


Administrative tools

Any web server infrastructure requires the existence of administrative tools to maintain and update the information used by the application. This information can include static content (web pages, graphic files), application source code, user authentication databases, etc. Depending on the site, technology, or software used, administrative tools will differ. For example, some web servers will be managed using administrative interfaces which are, themselves, web servers (such as the iPlanet web server) or will be administrated by plain text configuration files (in the Apache case[2]) or use operating-system GUI tools (when using Microsoft’s IIS server or ASP.Net).


In most cases, however, the server configuration will be handled using different file maintenance tools used by the web server, which are managed through FTP servers, WebDAV, network file systems (NFS, CIFS) or other mechanisms. Obviously, the operating system of the elements that make up the application architecture will also be managed using other tools. Applications may also have administrative interfaces embedded in them that are used to manage the application data itself (users, content, etc.).


Creating a map of the administrative interfaces used to manage the different parts of the architecture is very important, since if an attacker gains access to any of them he can then compromise or damage the application architecture. Thus in this phase it is important to:

  • List all the possible administrative interfaces.
  • Determine if administrative interfaces are available from an internal network or are also available from the Internet.


Some companies choose not to manage all aspects of their web server applications, but may have other parties managing the content delivered by the web application. This external company might either provide only parts of the content (news updates or promotions) or might manage the web server completely (including content and code). It is common to find administrative interfaces available from the Internet in these situations, since using the Internet is cheaper than providing a dedicated line that will connect the external company to the application infrastructure through a management-only interface. In this situation, it is very important to test if the administrative interfaces can be vulnerable to attacks.


References

  • [1] WebSEAL, also known as Tivoli Authentication Manager, is a reverse proxy from IBM which is part of the Tivoli framework.
  • [2] There are some GUI-based administration tools for Apache (like NetLoony) but they are not in widespread use yet.