Sunday, January 30, 2011

Exclusive: The 2011 DPI CEO Panel

To celebrate my blog’s first anniversary, I invited the CEOs of the 3 leading pure-player DPI vendors to participate in a Q&A panel.

My questions were submitted on Sunday, January 23rd to Rami Hadar CEO of Allot Communications (RH), Dave Caputo CEO of Sandvine Corp (DC) and James Brear CEO of Procera Networks (JB). All answers were received on Friday, January 28th and presented below in order received.

Since all 3 vendors are public companies, none of the questions was referring directly to the vendor’s business, to avoid un-necessary disclosures. Still, some questions were not answered by the vendors, due to their information disclosure polices.

What is your estimation for the DPI market size, globally?  
Rami Hadar, Allot

[RH] Hard to judge due to the fact that not all vendors put out figures. According to analyst reports we have seen, the market will do around $600-800M in 2011. On the customer side we continue to see strong demand.

[DC] 2010 - $400M 2011 - May well be dubbed “The Year of Network Policy Control” where Subscriber Quality of Experience and ARPU expansion will become the raison d'être for consumer internet service providers.

[JB] 2010 - $600M 2011 – $800M, (Service Provider DPI only)

What is the expected verticals share in revenues for 2011? 

[DC] Cable – 25% DSL - 50% Mobile - 22% Other Wireless -3%

[JB] 45% Cable 40% Mobile 15% Enterprise 

Which regions (NA, LAT, EMEA or Asia Pac) are expected to increase their share in DPI revenues during 2011?

[DC] AsiaPac, EMEA (both emerging markets Africa, Eastern Europe, as well as Western Europe)


Do you expect to see M&A activity in the DPI market in 2011-12? Of what kind?

[RH] The PCC (Policy Control & Charging) market is very hot at the moment. There are many small innovative companies involved in DPI, policy, charging and VAS that all have potentially good synergies. There are also a few large equipment manufactures that still do not have all the pieces of the puzzles – so M&As in and around our industry is a distinct possibility
Dave Caputo, Sandvine

[DC] Consolidation of PCRF providers, and acquisition of value-added-service vendors into broader equipment provider portfolios.

[JB] Yes. DPI companies will acquire related companies, and TEMs and other networking equipment vendors will pick up DPI companies. This is a natural development as the market is going through a transition stage (see below).

What is the expected impact of the new Net Neutrality rules?

[RH] The latest round in the NN debate in North America has opened up the doors for North American wireless operators to begin to immerse themselves in our technology and solutions and start playing catch up with the international peers.

[DC] New Network Neutrality rules help to stabilize the market, and are aligned with our objectives of transparent and reasonable network management. (refer to our blog entry)

[JB] Net Neutrality will continue to have some impact on service providers’ willingness to make investment decisions and to fully utilize the potential of policy enforcement. This will be the case until clear guidance is provided on what’s allowed or not.

What are the current popular use cases for DPI in mobile networks? Do they (solutions or customers) require integration with a standard PCRF?

[RH] Most of the use cases we are seeing do require integration with PCRF and in some cases OCS too. Some of the more popular are:

· Application-based tiered services

· Tier upsell/top-up

· Bill shock avoidance/advice of charge (roaming, tethering, other)

· Intelligent Charging

· Cell congestion management

[DC] Traffic visibility is still a high-runner with increased interest in analytics for proactive marketing/operations decision making. Next is usage management for differentiated service plans (based on combination of speed, quota, or type of usage). Price certainty is appealing to customers, e.g. they would pay more for unlimited social networking vs. a byte-based plan. Combining policy control and video optimization techniques supports such differentiated service plans. These use-cases do not require integration with standard PCRF, as the business rules and policy decision and enforcement can all be managed within integrated Sandvine system (although integration with PCRF is supported, if that is the service provider’s preferred deployment configuration).

[JB] Tiered services, business intelligence, and congestion management. Yes, the requirements require a PCRF – especially to accommodate future ambitions.

What are the current popular use cases for DPI in fixed networks? Do you see demand for a PCRF-like (i.e. standard, multi-device) control there as well?

[RH] Beyond traditional traffic management and optimization solutions we are seeing increased interest in DPI-based services such as media caching (Video, P2P, etc). Operators are looking for an Integrated solution – e.g. Caching and DPI on a single network element. PCRF-type solutions are just beginning to garner interest in the fixed market.

[DC] Traffic visibility is still a high-runner with increased interest in analytics for proactive marketing/operations decision making. Next is fairshare traffic management for fair use allocation of resources to address peak-time congestion issues. Use-cases are typically addressed with already-deployed network elements that provide same use case functionality (distributed policy enforcement function) as a PCRF so no pressing need to spend CAPEX on a new network element.

[JB] Business intelligence, network quality optimization, and added-value services. PCRF-like functionality will be required in some cases. Fixed operators are in general eyeing what mobile operators are doing.

DPI vendors are announcing an “eco-system” vision and are adding Value-added Services on top of the basic traffic management offering. Which of these services do you expect to succeed?

[RH] Media Optimization & Caching and different.Security services – such as protecting the network and subscriber from malicious traffic.

[DC] Content caching has been around for a number of years, and provides value both in terms of decreased latency for subscriber, as well as reduced transit costs for operator. Parental controls is also a solid VAS. Video and media optimization is gaining momentum, driven by streaming applications for mobile devices – operators are optimizing their network bandwidth. Mobile backhaul offload is a new area in which policy control will play a large role.

James Brear, Procera

[JB] DPI has evolved from being an autonomous system, doing P2P management, to being fully integrated into the PCC ecosystem. As such it offers a plethora of new functionality and values. DPI’s value propositions are intelligence, optimization, protection and tiered services. The latter is finally being realized which makes DPI more business critical, adds new values, and equals more customer-specific requirements to manage their unique value proposition and positioning in a highly competitive market.

How do you see the future of standalone DPI devices? How they will win against DPI embedded in existing network elements? (edge routers, GGSN)

[RH] We are obviously betting on the standalone side, for few reasons in particular:

· Diverting the resources of routers, switches, GGSNs, etc., to DPI functions impedes performance of their primary task

· Standalone DPI Platforms are access technology agnostic – Having a unified policy control and charging infrastructure in converged networks support the standalone approach.

· Intelligent (DPI based) Steering and LoadBalancing is becoming a must have element for enabling different VAS in broadband networks. Most of the existing network elements do not support these function.

[DC] Standalone equipment provides the superior technology, scale and reporting capabilities needed for networks that have multi-vendor access equipment and require common policy control. Although “embedded DPI” can provide some basic capabilities, it is limited in its protocol identification and reporting detail, and reduces the performance of its core functional requirement –session termination. Once “embedded DPI” is turned on, the overall session performance of that network node decreases significantly, at significant cost to the operator (higher than the cost standalone DPI).

[JB] Standalone and integrated will be complementary of each other and live in symbiosis. The architecture of DPI is not the same as a router or a switch, which means that some functionality will not fit naturally in such equipment. Customers also request more functionality to be incorporated into the DPI, which in many cases will require a dedicated hardware to process it. But we will certainly see an increasing amount of integrated DPI. This is inline with our “DPI Everywhere” vision, and as a software company we’re well positioned for this evolution.

When do you expect to see a new level of performance in DPI devices? What are the important performance parameters?

[RH] Performance is one of the main challenges in our industry. We are proud to be the first vendor that crossed the 100G barrier in our sector. We just released the next generation of our industry-leading Allot Service Gateway, the SG Sigma E. It boasts 160Gbps per platform, support for 8M subscribers and it can be clustered to provide 1Terrabit/s of throughput.

[DC] KPIs such as #concurrent subscribers, #concurrent flows, and #new flows/sec are all the more important in mobile environments with dynamic session setups. Others include Gbps throughput, and breadth of protocol signature recognition, accuracy of subscriber-usage mapping. Carrier-grade availability is crucial as policy control becomes integrated with billing systems.

[JB] There’s a constant drive for more and more capacity. However, right now customers are more focused on price-performance and getting best possible value for money. The natural next step will be DPI with 40 and/or 100 GE interfaces. This will go hand-in-hand with network design shifts, e.g. LTE gateways being deployed with these interfaces. It’s reasonable to expect a customer demand for this, and products that respond to such a demand, during this year (2011).


  1. Congratulation for the excellent work.
    IMO this blog is certainly among the top best sources, if not the best, for intelligence about the quickly evolving and imminently booming Broadband Traffic Management market and related M&A landscape.
    Continue the great work!

  2. I believe the M&A question/answer is interesting. I think 2011-2012 is when you'll finally see consolidation.

    Great work!

  3. How will companies like Boeing's Narus tech. fit into this picture? Will they hurt public companies because of their association with the heavy handed tactics in Egypt and other autocratic nations?

  4. Very well done. Absolutely brilliant information. I'm in love with this blog. they always provide such a great information. Safety Triangle NZ