Applying Filtering Services


This chapter describes ways to filter web traffic to reduce security risks or prevent inappropriate use. This chapter contains the following sections:

Filtering Overview

Filtering ActiveX Objects

Filtering Java Applets

Filtering URLs and FTP Requests with an External Server

Viewing Filtering Statistics and Configuration

Filtering Overview

This section describes how filtering can provide greater control over traffic passing through the security appliance. Filtering can be used in two distinct ways:

Filtering ActiveX objects or Java applets

Filtering with an external filtering server

Instead of blocking access altogether, you can remove specific undesirable objects from HTTP traffic, such as ActiveX objects or Java applets, that may pose a security threat in certain situations.

You can also use URL filtering to direct specific traffic to an external filtering server, such an Secure Computing SmartFilter (formerly N2H2) or Websense filtering server. Long URL, HTTPS, and FTP filtering can now be enabled using both Websense and Secure Computing SmartFilter for URL filtering. Filtering servers can block traffic to specific sites or types of sites, as specified by the security policy.


Note URL caching will only work if the version of the URL server software from the URL server vender supports it.


Because URL filtering is CPU-intensive, using an external filtering server ensures that the throughput of other traffic is not affected. However, depending on the speed of your network and the capacity of your URL filtering server, the time required for the initial connection may be noticeably slower when filtering traffic with an external filtering server.

Filtering ActiveX Objects

This section describes how to apply filtering to remove ActiveX objects from HTTP traffic passing through the firewall. This section includes the following topics:

ActiveX Filtering Overview

Enabling ActiveX Filtering

ActiveX Filtering Overview

ActiveX objects may pose security risks because they can contain code intended to attack hosts and servers on a protected network. You can disable ActiveX objects with ActiveX filtering.

ActiveX controls, formerly known as OLE or OCX controls, are components you can insert in a web page or other application. These controls include custom forms, calendars, or any of the extensive third-party forms for gathering or displaying information. As a technology, ActiveX creates many potential problems for network clients including causing workstations to fail, introducing network security problems, or being used to attack servers.

The filter activex command blocks the HTML <object> commands by commenting them out within the HTML web page. ActiveX filtering of HTML files is performed by selectively replacing the <APPLET> and </APPLET> and <OBJECT CLASSID> and </OBJECT> tags with comments. Filtering of nested tags is supported by converting top-level tags to comments.


Caution This command also blocks any Java applets, image files, or multimedia objects that are embedded in object tags .

If the <object> or </object> HTML tags split across network packets or if the code in the tags is longer than the number of bytes in the MTU, security appliance cannot block the tag.

ActiveX blocking does not occur when users access an IP address referenced by the alias command or for WebVPN traffic.

Enabling ActiveX Filtering

This section describes how to remove ActiveX objects in HTTP traffic passing through the security appliance. To remove ActiveX objects, enter the following command in global configuration mode:

hostname(config)# filter activex port[-port] local_ip local_mask foreign_ip foreign_mask

To use this command, replace port with the TCP port to which filtering is applied. Typically, this is port 80, but other values are accepted. The http or url literal can be used for port 80. You can specify a range of ports by using a hyphen between the starting port number and the ending port number.

The local IP address and mask identify one or more internal hosts that are the source of the traffic to be filtered. The foreign address and mask specify the external destination of the traffic to be filtered.

You can set either address to 0.0.0.0 (or in shortened form, 0) to specify all hosts. You can use 0.0.0.0 for either mask (or in shortened form, 0) to specify all hosts.

The following example specifies that ActiveX objects are blocked on all outbound connections:

hostname(config)# filter activex 80 0 0 0 0

This command specifies that the ActiveX object blocking applies to web traffic on port 80 from any local host and for connections to any foreign host.

To remove the configuration, use the no form of the command, as in the following example:

hostname(config)# no filter activex 80 0 0 0 0

Filtering Java Applets

This section describes how to apply filtering to remove Java applets from HTTP traffic passing through the firewall. Java applets may pose security risks because they can contain code intended to attack hosts and servers on a protected network. You can remove Java applets with the filter java command.

The filter java command filters out Java applets that return to the security appliance from an outbound connection. The user still receives the HTML page, but the web page source for the applet is commented out so that the applet cannot execute. The filter java command does not filter WebVPN traffic.


Note Use the filter activex command to remove Java applets that are embedded in <object> tags.


To remove Java applets in HTTP traffic passing through the firewall, enter the following command in global configuration mode:

hostname(config)# filter java  port[-port] local_ip local_mask foreign_ip foreign_mask

To use this command, replace port with the TCP port to which filtering is applied. Typically, this is port 80, but other values are accepted. The http or url literal can be used for port 80. You can specify a range of ports by using a hyphen between the starting port number and the ending port number.

The local IP address and mask identify one or more internal hosts that are the source of the traffic to be filtered. The foreign address and mask specify the external destination of the traffic to be filtered.

You can set either address to 0.0.0.0 (or in shortened form, 0) to specify all hosts. You can use 0.0.0.0 for either mask (or in shortened form, 0) to specify all hosts.

You can set either address to 0.0.0.0 (or in shortened form, 0) to specify all hosts. You can use 0.0.0.0 for either mask (or in shortened form, 0) to specify all hosts.

The following example specifies that Java applets are blocked on all outbound connections:

hostname(config)# filter java 80 0 0 0 0

This command specifies that the Java applet blocking applies to web traffic on port 80 from any local host and for connections to any foreign host.

The following example blocks downloading of Java applets to a host on a protected network:

hostname(config)# filter java http 192.168.3.3 255.255.255.255 0 0

This command prevents host 192.168.3.3 from downloading Java applets.

To remove the configuration, use the no form of the command, as in the following example:

hostname(config)# no filter java http 192.168.3.3 255.255.255.255 0 0

Filtering URLs and FTP Requests with an External Server

This section describes how to filter URLs and FTP requests with an external server. This section includes the following topics:

URL Filtering Overview

Identifying the Filtering Server

Buffering the Content Server Response

Caching Server Addresses

Filtering HTTP URLs

Filtering HTTPS URLs

Filtering FTP Requests

URL Filtering Overview

You can apply filtering to connection requests originating from a more secure network to a less secure network. Although you can use ACLs to prevent outbound access to specific content servers, managing usage this way is difficult because of the size and dynamic nature of the Internet. You can simplify configuration and improve security appliance performance by using a separate server running one of the following Internet filtering products:

Websense Enterprise for filtering HTTP, HTTPS, and FTP.

Secure Computing SmartFilter (formerly N2H2) for filtering HTTP, HTTPS, FTP, and long URL filtering.


Note URL caching will only work if the version of the URL server software from the URL server vender supports it.


Although security appliance performance is less affected when using an external server, users may notice longer access times to websites or FTP servers when the filtering server is remote from the security appliance.

When filtering is enabled and a request for content is directed through the security appliance, the request is sent to the content server and to the filtering server at the same time. If the filtering server allows the connection, the security appliance forwards the response from the content server to the originating client. If the filtering server denies the connection, the security appliance drops the response and sends a message or return code indicating that the connection was not successful.

If user authentication is enabled on the security appliance, then the security appliance also sends the user name to the filtering server. The filtering server can use user-specific filtering settings or provide enhanced reporting regarding usage.

Identifying the Filtering Server

You can identify up to four filtering servers per context. The security appliance uses the servers in order until a server responds. You can only configure a single type of server (Websense or Secure Computing SmartFilter ) in your configuration.


Note You must add the filtering server before you can configure filtering for HTTP or HTTPS with the filter command. If you remove the filtering servers from the configuration, then all filter commands are also removed.


Identify the address of the filtering server using the url-server command:

For Websense:

hostname(config)# url-server (if_name) host local_ip [timeout seconds] [protocol TCP | UDP
version [1|4] [connections num_conns] ]

For Secure Computing SmartFilter (formerly N2H2):

hostname(config)# url-server (if_name) vendor {secure-computing | n2h2} host
<local_ip> [port <number>] [timeout <seconds>] [protocol {TCP [connections <number>]} | 
UDP]

where <if_name> is the name of the security appliance interface connected to the filtering server (the default is inside).

For the vendor {secure-computing | n2h2}, you can use `secure-computing as a vendor string, however, `n2h2' is acceptable for backward compatibility. When the configuration entries are generated, `secure-computing' is saved as the vendor string.

The host <local_ip> is the IP address of the URL filtering server.

The port <number> is the Secure Computing SmartFilter server port number of the filtering server; the security appliance also listens for UDP replies on this port.


Note The default port is 4005. This is the default port used by the Secure Computing SmartFilter server to communicate to the security appliance via TCP or UDP. For information on changing the default port, please refer to the Filtering by N2H2 Administrator's Guide.


The timeout <seconds> is the number of seconds the security appliance should keep trying to connect to the filtering server.

The connections <number> is the number of tries to attempt to make a connection between the host and server.

For example, to identify a single Websense filtering server, enter the following command:

hostname(config)# url-server (perimeter) host 10.0.1.1 protocol TCP version 4

This identifies a Websense filtering server with the IP address 10.0.1.1 on a perimeter interface of the security appliance.Version 4, which is enabled in this example, is recommended by Websense because it supports caching.

To identify redundant Secure Computing SmartFilter servers, enter the following commands:

hostname(config)# url-server (perimeter) vendor n2h2 host 10.0.1.1
hostname(config)# url-server (perimeter) vendor n2h2 host 10.0.1.2

This identifies two Sentian filtering servers, both on a perimeter interface of the security appliance.

Buffering the Content Server Response

When a user issues a request to connect to a content server, the security appliance sends the request to the content server and to the filtering server at the same time. If the filtering server does not respond before the content server, the server response is dropped. This delays the web server response from the point of view of the web client because the client must reissue the request.

By enabling the HTTP response buffer, replies from web content servers are buffered and the responses are forwarded to the requesting client if the filtering server allows the connection. This prevents the delay that might otherwise occur.

To configure buffering for responses to HTTP or FTP requests, perform the following steps:


Step 1 To enable buffering of responses for HTTP or FTP requests that are pending a response from the filtering server, enter the following command:

hostname(config)# url-block block block-buffer-limit

Replace block-buffer with the maximum number of HTTP responses that can be buffered while awaiting responses from the url-server.


Note Buffering URLs longer than 3072 bytes are not supported.


Step 2 To configure the maximum memory available for buffering pending URLs (and for buffering long URLs), enter the following command:

hostname(config)# url-block mempool-size memory-pool-size

Replace memory-pool-size with a value from 2 to 10240 for a maximum memory allocation of 2 KB to 10 MB.


Caching Server Addresses

After a user accesses a site, the filtering server can allow the security appliance to cache the server address for a certain amount of time, as long as every site hosted at the address is in a category that is permitted at all times. Then, when the user accesses the server again, or if another user accesses the server, the security appliance does not need to consult the filtering server again.


Note Requests for cached IP addresses are not passed to the filtering server and are not logged. As a result, this activity does not appear in any reports. You can accumulate Websense run logs before using the url-cache command.


Use the url-cache command if needed to improve throughput, as follows:

hostname(config)# url-cache dst |   src_dst size

Replace size with a value for the cache size within the range 1 to 128 (KB).

Use the dst keyword to cache entries based on the URL destination address. Select this mode if all users share the same URL filtering policy on the Websense server.

Use the src_dst keyword to cache entries based on both the source address initiating the URL request as well as the URL destination address. Select this mode if users do not share the same URL filtering policy on the Websense server.

Filtering HTTP URLs

This section describes how to configure HTTP filtering with an external filtering server. This section includes the following topics:

Configuring HTTP Filtering

Enabling Filtering of Long HTTP URLs

Truncating Long HTTP URLs

Exempting Traffic from Filtering

Configuring HTTP Filtering

You must identify and enable the URL filtering server before enabling HTTP filtering.

When the filtering server approves an HTTP connection request, the security appliance allows the reply from the web server to reach the originating client. If the filtering server denies the request, the security appliance redirects the user to a block page, indicating that access was denied.

To enable HTTP filtering, enter the following command:

hostname(config)# filter url [http | port[-port] local_ip local_mask foreign_ip 
foreign_mask] [allow] [proxy-block] 

Replace port with one or more port numbers if a different port than the default port for HTTP (80) is used. Replace local_ip and local_mask with the IP address and subnet mask of a user or subnetwork making requests. Replace foreign_ip and foreign_mask with the IP address and subnet mask of a server or subnetwork responding to requests.

The allow option causes the security appliance to forward HTTP traffic without filtering when the primary filtering server is unavailable. Use the proxy-block command to drop all requests to proxy servers.

Enabling Filtering of Long HTTP URLs

By default, the security appliance considers an HTTP URL to be a long URL if it is greater than 1159 characters. You can increase the maximum length allowed.

Configure the maximum size of a single URL with the following command:

hostname(config)# url-block url-size long-url-size

Replace long-url-size with the maximum size in KB for each long URL being buffered. For Websense, this is a value from 2 to 4 for a maximum URL size of 2 KB to 4 KB; for Secure Computing, this is a value between 2 to 3 for a maximum URL size of 2 KB to 3 KB. The default value is 2.

Truncating Long HTTP URLs

By default, if a URL exceeds the maximum permitted size, then it is dropped. To avoid this, you can set the security appliance to truncate a long URL by entering the following command:

hostname(config)# filter url [longurl-truncate | longurl-deny | cgi-truncate]

The longurl-truncate option causes the security appliance to send only the hostname or IP address portion of the URL for evaluation to the filtering server when the URL is longer than the maximum length permitted. Use the longurl-deny option to deny outbound URL traffic if the URL is longer than the maximum permitted.

Use the cgi-truncate option to truncate CGI URLs to include only the CGI script location and the script name without any parameters. Many long HTTP requests are CGI requests. If the parameters list is very long, waiting and sending the complete CGI request including the parameter list can use up memory resources and affect firewall performance.

Exempting Traffic from Filtering

To exempt specific traffic from filtering, enter the following command:

hostname(config)# filter url except source_ip source_mask dest_ip dest_mask 

For example, the following commands cause all HTTP requests to be forwarded to the filtering server except for those from 10.0.2.54.

hostname(config)# filter url http 0 0 0 0
hostname(config)# filter url except 10.0.2.54 255.255.255.255 0 0

Filtering HTTPS URLs

You must identify and enable the URL filtering server before enabling HTTPS filtering.


Note Websense and Smartfilter currently support HTTPS; older versions of Secure Computing SmartFilter (formerly N2H2) did not support HTTPS filtering.


Because HTTPS content is encrypted, the security appliance sends the URL lookup without directory and filename information. When the filtering server approves an HTTPS connection request, the security appliance allows the completion of SSL connection negotiation and allows the reply from the web server to reach the originating client. If the filtering server denies the request, the security appliance prevents the completion of SSL connection negotiation. The browser displays an error message such as "The Page or the content cannot be displayed."


Note The security appliance does not provide an authentication prompt for HTTPS, so a user must authenticate with the security appliance using HTTP or FTP before accessing HTTPS servers.


To enable HTTPS filtering, enter the following command:

hostname(config)# filter https port[-port] localIP local_mask foreign_IP foreign_mask 
[allow]

Replace port[-port] with a range of port numbers if a different port than the default port for HTTPS (443) is used.

Replace local_ip and local_mask with the IP address and subnet mask of a user or subnetwork making requests.

Replace foreign_ip and foreign_mask with the IP address and subnet mask of a server or subnetwork responding to requests.

The allow option causes the security appliance to forward HTTPS traffic without filtering when the primary filtering server is unavailable.

Filtering FTP Requests

You must identify and enable the URL filtering server before enabling FTP filtering.


Note Websense and Smartfilter currently support FTP; older versions of Secure Computing SmartFilter (formerly known as N2H2) did not support FTP filtering.


When the filtering server approves an FTP connection request, the security appliance allows the successful FTP return code to reach originating client. For example, a successful return code is "250: CWD command successful." If the filtering server denies the request, alters the FTP return code to show that the connection was denied. For example, the security appliance changes code 250 to "550 Requested file is prohibited by URL filtering policy."

To enable FTP filtering, enter the following command:

hostname(config)# filter ftp port[-port] localIP local_mask foreign_IP foreign_mask 
[allow] [interact-block]

Replace port[-port] with a range of port numbers if a different port than the default port for FTP (21) is used.

Replace local_ip and local_mask with the IP address and subnet mask of a user or subnetwork making requests.

Replace foreign_ip and foreign_mask with the IP address and subnet mask of a server or subnetwork responding to requests.

The allow option causes the security appliance to forward HTTPS traffic without filtering when the primary filtering server is unavailable.

Use the interact-block option to prevent interactive FTP sessions that do not provide the entire directory path. An interactive FTP client allows the user to change directories without typing the entire path. For example, the user might enter cd ./files instead of cd /public/files.

Viewing Filtering Statistics and Configuration

This section describes how to monitor filtering statistics. This section includes the following topics:

Viewing Filtering Server Statistics

Viewing Buffer Configuration and Statistics

Viewing Caching Statistics

Viewing Filtering Performance Statistics

Viewing Filtering Configuration

Viewing Filtering Server Statistics

To show information about the filtering server, enter the following command:

hostname# show running-config url-server

The following is sample output from the show running-config url-server command:

hostname# show running-config url-server
url-server (outside) vendor n2h2 host 128.107.254.202 port 4005 timeout 5 protocol TCP

To show information about the filtering server or to show statistics, enter the following command:

The following is sample output from the show running-config url-server statistics command, which shows filtering statistics:

hostname# show running-config url-server statistics

Global Statistics:
--------------------
URLs total/allowed/denied         13/3/10
URLs allowed by cache/server      0/3
URLs denied by cache/server       0/10
HTTPSs total/allowed/denied       138/137/1
HTTPSs allowed by cache/server    0/137
HTTPSs denied by cache/server     0/1
FTPs total/allowed/denied         0/0/0
FTPs allowed by cache/server      0/0
FTPs denied by cache/server       0/0
Requests dropped                  0
Server timeouts/retries           0/0
Processed rate average 60s/300s   0/0 requests/second
Denied rate average 60s/300s      0/0 requests/second
Dropped rate average 60s/300s     0/0 requests/second

Server Statistics:
--------------------
10.125.76.20                      UP
  Vendor                          websense
  Port                            15868
  Requests total/allowed/denied   151/140/11
  Server timeouts/retries         0/0
  Responses received              151
  Response time average 60s/300s  0/0

URL Packets Sent and Received Stats:
------------------------------------
Message                 Sent    Received
STATUS_REQUEST          1609    1601
LOOKUP_REQUEST          1526    1526
LOG_REQUEST             0       NA

Errors:
-------
RFC noncompliant GET method     0
URL buffer update failure       0

Viewing Buffer Configuration and Statistics

The show running-config url-block command displays the number of packets held in the url-block buffer and the number (if any) dropped due to exceeding the buffer limit or retransmission.

The following is sample output from the show running-config url-block command:

hostname# show running-config url-block
    url-block url-mempool 128 
    url-block url-size 4 
    url-block block 128 

This shows the configuration of the URL block buffer.

The following is sample output from the show url-block block statistics command:

hostname# show running-config url-block block statistics

URL Pending Packet Buffer Stats with max block  128 
----------------------------------------------------- 
Cumulative number of packets held:              896
Maximum number of packets held (per URL):       3
Current number of packets held (global):        38
Packets dropped due to
       exceeding url-block buffer limit:        7546
       HTTP server retransmission:              10
Number of packets released back to client:      0

This shows the URL block statistics.

Viewing Caching Statistics

The following is sample output from the show url-cache stats command:

hostname# show url-cache stats
URL Filter Cache Stats
----------------------
    Size :	     128KB
 Entries :	      1724
  In Use :	       456
 Lookups :	        45
    Hits :	         8

This shows how the cache is used.

Viewing Filtering Performance Statistics

The following is sample output from the show perfmon command:

hostname# show perfmon
PERFMON STATS:    Current      Average
Xlates               0/s          0/s
Connections          0/s          2/s
TCP Conns            0/s          2/s
UDP Conns            0/s          0/s
URL Access           0/s          2/s
URL Server Req       0/s          3/s
TCP Fixup            0/s          0/s
TCPIntercept         0/s          0/s
HTTP Fixup           0/s          3/s
FTP Fixup            0/s          0/s
AAA Authen           0/s          0/s
AAA Author           0/s          0/s
AAA Account          0/s          0/s

This shows URL filtering performance statistics, along with other performance statistics. The filtering statistics are shown in the URL Access and URL Server Req rows.

Viewing Filtering Configuration

The following is sample output from the show running-config filter command:

hostname# show running-config filter
filter url http 0.0.0.0 0.0.0.0 0.0.0.0 0.0.0.0