Online Behavioral Targeting is an online advertising approach that involves presenting users with advertisements based on their interests, as deduced by monitoring their web browsing preferences. The Cisco Service Control Engine (Cisco SCE) platform can enable online behavioral targeting based on an analysis of subscriber online usage patterns.
Such behavioral targeting does not require the analysis of each and every HTTP request on the line, because such analysis would result in numerous excess information. The Cisco SCE platform performs the first level of analysis in the behavioral targeting chain by inspecting the user browsing sessions, detecting the particular requests that are triggered by the actual user browsing (these events are termed ClickStream), and generating Raw Data Records (RDRs) that contain a digest of these events. To avoid compromising subscriber privacy, the RDRs can be configured to not include any Personally Identifiable Information (PII), in which case all elements containing PII are hashed. The RDRs are typically received by an entity that analyzes the nature of usage and creates a profile of the subscriber to be used later for targeting. The way the greater solution works is outside the scope of this document.
Figure 1 illustrates the high-level overview of an RDR-based behavioral targeting solution.
Figure 1 High Level Overview of an RDR-based Behavioral Targeting Solution
ClickStream detection is a fundamental capability of the solution, because it can detect which specific requests, out of the enormous number of HTTP requests generated throughout the subscriber web activity, were actually triggered by the subscriber browsing the web. When a subscriber clicks a link, or enters a URL in the browser address bar, an HTTP request is generated to fetch this URL. Typically, an HTML page is returned, which constitutes the outline of the contents requested. For the browser to be able to render this page, it must download multiple objects (tens or sometimes around a hundred for a single page viewed), which in turn results in multiple HTTP requests for obtaining these objects.
To carry out behavioral targeting, it is sufficient to understand what the user was trying to do (represented by the initial request, such as biz.publisher.com/ap/081120/world_markets.html > global markets), rather than looking at each object downloaded as a secondary result of such a request (such as: http://ads.adnetwork.com/a/a/in/interbroke/300x250_yah.jpg > broker advertisement).
ClickStream detection makes exactly this distinction, allowing the number of requests to be analyzed to be greatly reduced, which is necessary to enable a scalable analysis solution.
The information that is collected per such transaction is exported from the Cisco SCE using Extended Transaction Usage RDRs, which includes information on the transaction that has been performed, in addition to information on the subscriber that performed the transaction.
PII, such as subscriber ID and IP address, is protected by hashing this information with the use of preconfigured “salt”. In this way, the subscriber record can be matched only by another system configured with the same salt, in which case it has the original PII to match against. Salt is configurable to the Cisco SCE platform.
ClickStream signatures are mapped by default to the HTTP Browsing protocol and consequently to the browsing service. To be able to act on them separately, first move them to a protocol of their own, then assign this protocol to a service of its own.
Figure 2 and Figure 3 illustrate configuring the ClickStream protocol and service.
Figure 2 Configuring the ClickStream Protocol
Figure 3 Configuring the ClickStream Service
Configuring Extended Transaction Usage RDRs
Extended Transaction Usage RDRs (TURs) contain extra protocol-specific information (see the “Anonymized HTTP Transaction Usage RDR” section). These RDRs are configured from the Transaction Usage RDRs tab of the RDR Settings dialog box of the Cisco Service Control Application for Broadband (Cisco SCA BB) console.
Enable these RDRs explicitly by checking the appropriate check boxes, and choosing the packages you want them generated for (you only need to enable these RDRs for the ClickStream service in each package).
Optionally, these RDRs can be anonymized, in which case all PII in the RDR is hashed using Message Digest 5 (MD5). Hashing is done using the “salt” configured on the SCE platform. For GUI options on anonymizing the RDRs, see Transaction Usage RDR Settings (Figure 4).
Figure 4 Transaction Usage RDR Settings
Enabling Deep HTTP Inspection
To ensure comprehensive detection of the ClickStream events in the traffic stream, it is important to enable deep inspection of HTTP, which configures the Cisco SCE platform to analyze and classify all HTTP requests within a single flow.
Some browsers, in conjunction with some web server implementations, use the same TCP flow to carry multiple requests triggered by clicks that target the same host. Such events are not detected if the classification is done only at the beginning of the flow (which is the default for Cisco SCA BB).
To enable deep HTTP inspection, in the Cisco SCA BB Console Service Configuration Editor, choose Configuration > System Settings > Advanced Options > Advanced Service Configuration Options.
Note Enabling deep HTTP inspection is expected to impact the Cisco SCE performance because of the excessive processing associated with it, the actual figure depending on the amount and on the nature of HTTP traffic. We recommend that you monitor SCE platform performance when enabling this capability.
Configuring MD5 Salt to the SCE Platform
When the Anonymize option is selected, Extended TURs are generated with all personally identifying fields (subscriber ID and subscriber IP address) hashed using MD5.
The Cisco SCE platform applies a “salt” to the personally identifying field before hashing it. The salt is 128 bits (16 bytes) long, and it is configured to the SCE platform in four separate 4-byte arguments represented in HEX. The default value of the salt is 0x12345678 0x12345678 0x12345678 0x12345678.
The salt is configured to the SCE by using these CLI commands with four 4-byte arguments in HEX:
SCE10000(config)# interface linecard 0
SCE10000(config if)# salt 0x12345678 0x00004321 0xfafafafa 0xafafafaf
To return the salt to the default value use these CLI commands:
SCE10000(config)# interface linecard 0
SCE10000(config if)# default salt
Save the running configuration by using this command:
SCE10000# copy running-config startup-config
Configuring RDR Routing to the Profiling Server
Ensure that only ClickStream RDRs are sent to the designated server, even if other RDRs are enabled on the system. This is done by directing the extended HTTP TUR to a separate category, and routing this category to a dedicated server.
This is accomplished in two steps:
1. Direct the HTTP Extended TUR to an exclusive RDR category.
Map the Extended TUR for HTTP (tag 0xF0F0F53C / 4042323260) to RDR category 2 by using this CLI command:
Step 10 In the Cisco SCA BB Policy Editor, click the Classification tab (left pane), and highlight the Browsing service.
Step 11 Click the Add (+) icon to add a new service under the Browsing service.
Step 12 Name the service ClickStream (or any other name you choose) (see Figure 8).
Figure 8 Service Settings
Step 13 Click the Hierarchy tab (see Figure 9) and check the two check boxes to add a dedicated service counter to the ClickStream Service. (This is useful if you want to generate reports on the global and per-subscriber ClickStream activity.)
Figure 9 Hierarchy
Step 14 Click OK.
Step 15 In the right pane, click the Add (+) icon to add a service element.
Step 16 In the dialog box that opens, click Select next to the Protocol field and select the ClickStream Event protocol (or whatever you named your ClickStream protocol) from the list (see Figure 10).
Step 17 Click OK.
Figure 10 Edit Service Element
Step 18In the Cisco SCA BB Policy Editor, click the Policies tab (left pane), and then select the package for which to generate the ClickStream RDRs.
Step 19 In the right pane, click the Add (+) icon to add the ClickStream service.
Step 20 In the window that opens (see Figure 11), select ClickStream from the drop-down list.
Figure 11 Add New Rule to Package
Step 21Click OK.
Step 22 In the Cisco SCA BB Policy Editor, repeat Step 18 through Step 21 for every package for which you want to enable ClickStream RDRs.
Step 23 In the right pane, click Configuration and choose Classification > RDR Settings.
Step 24 Click the Transaction Usage RDRs tab (see Figure 12).
Figure 12 RDR Settings—Transaction Usage RDRs
Step 25 Enable Extended TURs by checking the relevant check box.
Step 26 (Optional) Enable the anonymize option as needed, by checking Anonymize - Encrypt Personally Identifiable Information.
Step 27 Check the ClickStream service on all relevant packages, to enable generating RDRs for this service. Uncheck any other service for which you do not want Transaction Usage RDRs (extended or regular) generated.
Note Make sure that you have already created a specific rule for ClickStream on all relevant packages.
Step 28 Choose Policies > Configuration > System Settings (see Figure 13).
Figure 13 Service Configuration Editor—Policies > Configuration > System Settings
Step 29 On the Advanced Options tab, click Advanced Service Configuration Options (see Figure 14) to enable deep inspection of HTTP flows (beyond the first transaction) by setting the highlighted value to 64000. This enables the analysis of multiple transactions within a single HTTP flow, which is important for comprehensive detection of ClickStream events.
Figure 14 Advanced Service Configuration Options
Step 30 Apply the Service Configuration to the SCE platform.
Step 31 (Optional) If the anonymize option is enabled, configure MD5 salt to the SCE platform (use a 128-bit salt value of your choice) by using this command:
SCE10000(config if)# salt 0xfafafafa 0xfafafafa 0xfafafafa 0xfafafafa
Step 32 Map the Extended TUR for HTTP (tag 0xF0F0F53C / 4042323260) to RDR category 2 by using this command:
User agent field extracted from the HTTP transaction.
REFERER extracted from the HTTP transaction.
Cookie extracted from the HTTP transaction.
Further elaboration on RDR fields:
HASHED_SUBSCRIBER_ID—Subscriber identification string, introduced through the subscriber management interfaces, passed through hashing algorithm as described in the “Hash Algorithm” section.
The field is a 32-byte-long string, containing a Hexadecimal notation of the 128-bit hash result.
PACKAGE_ID—ID of the package assigned to the subscriber whose traffic is being reported. An assigned Package ID is an integer value between 0 and maximum_number_of_packages. The maximum_number_of_packages value is reserved for unknown subscribers.
HASHED_SUBSCRBIER_IP—IP address of the subscriber side of the reported session, after passing through the hashing algorithm as described in the “Hash Algorithm” section.
The field is a 32-byte-long string, containing a hexadecimal notation of the 128-bit hash result.
CLIENT_PORT—For TCP/UDP-based sessions, the port number of the client side (initiator) of the networking session. For non-TCP/UDP sessions, this field has the value 0.
CONFIGURED_DURATION—For periodic RDRs, the configured period, in seconds, between successive RDRs.
END_TIME—Ending time stamp of this RDR. The field is in UNIX time_t format, which is the number of seconds since midnight of January 1, 1970.
FLAVOR_ID—For protocol signatures that have flavors, this field contains the ID of the flavor associated with this session.
PROTOCOL_ID—Contains the unique ID of the protocol associated with the reported session.
Note The PROTOCOL_ID is the Generic IP, Generic TCP, or Generic UDP protocol ID value, according to the specific transport protocol of the transaction, unless a more specific protocol definition (such as a signature-based protocol or a port-based protocol), which matches the reported session, is assigned to a service.
PROTOCOL_SIGNATURE—Contains the ID of the protocol signature associated with this session.
REPORT_TIME—Ending time stamp of this RDR. The field is in UNIX time_t format, which is the number of seconds since midnight of January 1, 1970.
SERVER_IP—Contains the destination IP address of the reported session. (The destination is defined as the server or the listener of the networking session.) The IP address is in a 32-bit binary format.
SERVER_PORT—For TCP/UDP-based sessions, this field contains the destination port number of the networking session. For non-TCP/UDP sessions, this field contains the IP protocol number of the session flow.
SERVICE_ID—Indicates the service classification of the reported session. For example, in the Transaction RDR, this field indicates which service was accessed, and in the Breaching RDR, this field indicates which service was breached.
TIME_FRAME—System supports time-dependent policies, by using different rules for different time frames. This field indicates the time frame during which the RDR was generated. The value can be in the range from 0 to 3, indicating which of the four time frames was used.
ZONE_ID—Contains the ID of the zone associated with this session.
Note All volumes in RDRs are reported in Layer 3 bytes.
Hashing is done using the MD5 hash function, using a “salt” prepended to the hashed value.
“salt” means adding a predefined value before the to-be hashed fields, to avoid simple backtracking of the source argument. The salt currently used by the SCE platform is 128 bits long and is configured to the SCE in four separate 4-byte arguments represented in HEX.
Setting the salt is done through CLI using a hexadecimal notation. It can be updated periodically. (By default, it set to 0x12345678 0x12345678 0x12345678 0x12345678.)
Hash result is 128 bits long, and is represented in RDRs as a 32-byte string. Each byte represents 4 bits of the result. To make it printable, add 64. (for example, 0000 mapped to ASCII(64) = '@'; 1111 mapped to ASCII(127) = '_').
Short-hash that appears in RDRs is composed of the lower 32 bits of the hash result.
Obtaining Documentation and Submitting a Service Request
Subscribe to What’s New in Cisco Product Documentation, which lists all new and revised Cisco technical documentation, as an RSS feed and deliver content directly to your desktop using a reader application. The RSS feeds are a free service.
Cisco and the Cisco logo are trademarks or registered trademarks of Cisco and/or its affiliates in the U.S. and other countries. To view a list of Cisco trademarks, go to this URL: www.cisco.com/go/trademarks. Third-party trademarks mentioned are the property of their respective owners. The use of the word partner does not imply a partnership relationship between Cisco and any other company. (1110R)
Any Internet Protocol (IP) addresses used in this document are not intended to be actual addresses. Any examples, command display output, and figures included in the document are shown for illustrative purposes only. Any use of actual IP addresses in illustrative content is unintentional and coincidental.