Understanding the demand for deep packet inspection

Editor’s note: RCR Wireless News has teamed up with Yankee Group to conduct a series of market surveys of RCR Wireless News’ subscribers to gauge their thoughts on various technology issues impacting the wireless ecosystem. RCR Wireless News will publish the exclusive results from our joint project, with Yankee Group’s expert analysis, both in our print issue and on our Web site.
More than 2,200 subscribers responded to the first wave of the study. Those of you who respond to the surveys will get an executive summary of the survey results and are eligible to win an American Express gift certificate.
The second set of surveys is scheduled to launch in early June and many of you will be contacted by e-mail. We encourage your participation to ensure the continued high quality of survey results and to make the data meaningful to you and your peers.
While not without controversy, the market for deep packet inspection (DPI) solutions among communications service providers has grown precipitously over the past 18 to 24 months. The technology has been around for a number of years, but service provider awareness and adoption had been slow as many providers were unfamiliar with the technology or unconvinced of its benefits. That is no longer the case, particularly among mobile operators. While DPI has seen most of its growth to date in the cable and DSL markets, mobile carriers represent the next fertile frontier for the technology.
In a recent survey of network operators and other telecommunications industry professionals conducted by RCR Wireless News and Yankee Group, 57% of respondents described themselves as either very familiar or somewhat familiar with DPI and its benefits. Even more telling, 77% of respondents were either strongly interested or somewhat interested in DPI, indicating that operator demand is even outpacing a clear understanding of what DPI brings to the table.
Given that there is now an understood role in the network for DPI, the more salient questions are what manner are operators primarily interested in deploying and what are the primary benefits they are looking to achieve. Previously, there has been debate within the market over whether DPI should be deployed passively in the network or in an active role within the routing path. Operator survey data clearly indicates that this debate has passed. Only 5% of respondents indicate that they would prefer an off-line, passive solution, while 70% of respondents would opt for a solution that would need to be placed in-line (see chart 1).
This is indicative of a number of things. First, operators are looking to DPI today for more than just passive monitoring and network visibility, as will be discussed later. More than that, there is a comfort level with the technology that had not previously existed, and this mitigates operator reticence to place it actively within the routing path. And finally, DPI solutions have matured to a degree (primarily in regards to increased available throughput rates that now typically exceed 10 gigabits per second full-duplex) that such a deployment would not cause significant latency or delay on the network.
Yet this does not indicate that as DPI solutions migrate to almost in-line deployments, those deployments will look the same as those that are deployed passively. Typically DPI has been implemented as a standalone, purpose-built solution at key points on the network. However, as many operators accept a broader vision for what the technology can do, there is a growing sense that DPI functionality should be folded into other distinct network elements. This too is evident in Exhibit 1, where roughly two-thirds of those that preferred in-line deployments would like to see DPI monitoring and control capabilities embedded into another network element to broaden its usefulness and effectiveness. This is particularly true among the tier-one operator community, where there is increasing demand for DPI-level intelligence in a larger package. The reason for this is two-fold. First, it allows the inspection technology to easily feed intelligence into other control mechanisms, and thus increases the ability to monetize that intelligence. Secondly, it decreases the number of distinct elements that an operator must place directly in the routing path, therefore reducing network complexity and potential points of failure or delay.
Upon establishing this market demand, the question becomes what other network elements present the most compelling cases for integration with DPI. Many operators and vendors are looking to DPI for security applications, particularly in networks designed for delivering enterprise services. Simultaneously, DPI is seeing increased usage in conjunction with policy management and traffic shaping technologies. These sorts of deployments are designed around service control rather than pure service monitoring, as operators use DPI intelligence to inform intelligent provisioning of the network and implementation of subscriber or application-specific policies. IP/MPLS vendors are looking to fold DPI into large aggregation routers for similar purposes, where pure router capacity is augmented by better informed network policies. And further out on the horizon, as operators look to engage in targeted advertising to augment subscriber revenue streams, DPI has a role to play in conjunction with ad-insertion technologies to capitalize on granular subscriber preference data.
All of these related technologies (as well as others such as session border control) represent legitimate directions for the DPI market. As of now, carrier security and policy management directed deployments are the most in-line with carrier demands (see chart 2).
Yet what operators are indicating is that unique needs and goals for the technology will continue to drive DPI in a variety of distinct directions. Therefore, the vendor community would be wise to focus its strategic direction on moving away from the standalone model. Openly architected solutions and robust, diverse partner ecosystems will be the best ways to capitalize on the various pockets of demand. Rather than developing product packages that combine DPI with other solutions such as policy management, security or online charging on a one-off basis to respond to specific service provider requests, vendors need to develop pre-packaged offerings that address the broader visions of their customers.
The final key question for this market is what are these broader visions. Armed with in-line, pre-integrated solutions capable of not just monitoring, but affecting, traffic, what do operators want to do with the technology? Chart 3 illustrates some of these key uses.
What this indicates is that while many operators describe all of these factors as “very important” in driving their DPI investment, the most important uses today are around internally facing network optimization and control. Solutions designed for security detection and mitigation and quality of service prioritization, while not readily transparent to the subscriber base, allow operators to maximize network efficiency and ultimately maintain subscriber happiness. Recent outcries around traffic shaping and blocking techniques from service providers in the U.S. and Canada have helped to tamper down deployment plans aimed at either peer-to-peer traffic control or the introduction of new service packages. These sorts of deployments, while still relevant, are more outwardly facing and present public relations challenges that not all service providers are ready to face at this time.